Integrating volterra series model and deep neural networks to equalize nonlinear power amplifiers

Description

FIELD OF THE INVENTION

The present invention relates to the field of equalization of nonlinear radio frequency power amplifiers, and more particularly to a neural network implementation of radio frequency power amplifier equalization

BACKGROUND OF THE INVENTION

Most modern wireless communication systems, including the fifth generation (5G) cellular systems, use multi-carrier or OFDM (orthogonal frequency division multiplexing) modulations whose signals have extremely high peak to average power ratio (PAPR). This makes it challenging to enhance the efficiency of power amplifiers (PAs). Signals with high PAPR need linear power amplifier response in order to reduce distortion. Nevertheless, PAs have the optimal power efficiency only in the nonlinear saturated response region. In practice, the PAs in the wireless transceivers have to work with high output backoff (OBO) in order to suppress nonlinear distortions, which unfortunately results in severe reduction of power efficiency [1]. This problem, originated from the nonlinearity of PAs, has been one of the major constraints to enhance the power efficiency of modern communication systems.

Various strategies have been investigated to mitigate this problem. The first strategy is to reduce the PAPR of the transmitted signals. Many techniques have been developed for this purpose, such as signal clipping, peak cancellation, error waveform subtraction [2]. For OFDM signals, pilot tones and unmodulated subcarriers can be exploited to reduce PAPR with some special pre-coding techniques [3].

The second strategy is to linearize the PAs at the transmitters. One of the dominating practices today is to insert a digital pre-distorter (DPD) before the PA, which distorts the signals appropriately so as to compensate for the nonlinear PA response [4] [5] [6]. DPD has been applied widely in many modern transmitters [2].

The third strategy is to mitigate the nonlinear PA distortions at the receivers via post-distorter equalization [7] [8] [9]. The solution presented in [10] develops a Bayesian signal detection algorithm based on the nonlinear response of the PAs. However, this method works for the simple “AM-AM AM-PM” nonlinear PA model only. Alternatively, as a powerful nonlinear modeling tool, artificial neural networks have also been studied for both nonlinear modeling of PAs [11] [12] and nonlinear equalization [13] [14] [15].

One of the major design goals for the 5G systems is to make the communication systems more power efficient. This needs efficient PAs, which is unfortunately more challenging since 5G signals have much higher PAPR and wider bandwidth [16] [17]. This is an especially severe problem for cost and battery limited devices in massive machine-type communications or internet of things (IoT). Existing nonlinear PA mitigation strategies may not be sufficient enough. PAPR can be reduced to some extent only. DPD is too complex and costly for small and low-cost 5G devices. Existing DPD and equalization techniques have moderate nonlinear distortion compensation capabilities only.

As a matter of fact, the nonlinear equalization strategy is more attractive to massive MIMO and millimeter wave transmissions due to the large number of PAs needed [18] [19] [20]. Millimeter wave transmissions require much higher transmission power to compensate for severe signal attenuation. Considering the extremely high requirement on PA power efficiency and the large number of PAs in a transmitter, the current practice of using DPD may not be appropriate due to implementation complexity and cost.

There are various types of intermodulation that can be found in radio systems, see, Rec. ITU-R SM.1446: Type 1 Single channel intermodulation: where the wanted signal is distorted by virtue of non-linearities in the transmitter; Type 2 Multichannel intermodulation: where the wanted signals of multi channels are distorted by virtue of non-linearities in the same transmitter; Type 3 Inter transmitter intermodulation: where one or more transmitters on a site intermodulate, either within the transmitters themselves or within a non-linear component on site to produce intermodulation products; Type 4 Intermodulation due to active antennas: the multicarrier operating mode of an active antenna, along with the non-linearity of amplifiers, originates spurious emissions under the form of intermodulation signals; and Type 5 Intermodulation due to passive circuits: where transmitters share the same radiating element and intermodulation occurs due to non-linearities of passive circuits. See, Rep. ITU-R-SM.2021

An amplifier can be characterized by a Taylor series of the generalized transfer function [32]

i₀+k₁e_IN+k₂e_IN²+k₃e_IN³+k₄e_IN⁴+k₅e_IN⁵+ . . .

where i₀is the quiescent output current, k1, k2, etc. are coefficients and eIN represents the input signal. When two sinusoidal frequencies ω₁=2πƒ₁and ω₂=2πƒ₂of the amplitude a₁and a₂are applied to the input of the amplifier, the input signal is:

e_IN=a₁cos ω₁t+a₂cos ω₂t

and the output i_OUTmay be shown to be the sum of the DC components:

$i_{O U T} = i_{0} + \frac{k_{2}}{2} (a_{1}^{2} + a_{2}^{2}) + \frac{k_{4}}{8} (3 a_{1}^{4} + 1 2 a_{1}^{2} + 3 a_{2}^{4})$

fundamental components:

$+ (k_{1} a_{1} + \frac{3}{4} k_{3} a_{1}^{3} + \frac{3}{2} k_{3} a_{1} a_{2}^{2} + \frac{5}{8} k_{5} a_{1}^{5} + \frac{15}{4} k_{5} a_{1}^{3} a_{2}^{2} + \frac{15}{8} k_{5} a_{1} a_{2}^{4}) \cos ω_{1} t + (k_{1} a_{2} + \frac{3}{4} k_{3} a_{2}^{3} + \frac{3}{2} k_{3} a_{1}^{2} a_{2} + \frac{5}{8} k_{5} a_{2}^{5} + \frac{15}{4} k_{5} a_{1}^{2} a_{2}^{3} + \frac{15}{8} k_{5} a_{1}^{4} a_{2}) \cos ω_{2} t$

2nd order components:

$\begin{matrix} + (\frac{1}{2} k_{2} a_{1}^{2} + \frac{1}{2} k_{3} a_{1}^{4} + \frac{3}{2} k_{4} a_{1}^{2} a_{2}^{2}) \cos 2 ω_{1} t + (\frac{1}{2} k_{2} a_{2}^{2} + \frac{1}{2} k_{3} a_{2}^{4} + \frac{3}{2} k_{4} a_{1}^{2} a_{2}^{2}) \cos 2 ω_{2} t + (k_{2} a_{1} a_{2} + \frac{3}{2} k_{4} a_{1}^{3} a_{2} + \frac{3}{2} k_{4} a_{1} a_{2}^{3}) \cos (ω_{1} \pm ω_{2}) t \end{matrix}$

3rd order components:

$+ (\frac{1}{4} k_{3} a_{1}^{3} + \frac{5}{1 6} k_{5} a_{1}^{5} + \frac{5}{4} k_{5} a_{1}^{3} a_{2}^{2}) \cos 3 ω_{1} t + (\frac{1}{4} k_{3} a_{2}^{3} + \frac{5}{1 6} k_{5} a_{2}^{5} + \frac{5}{4} k_{5} a_{1}^{2} a_{2}^{3}) \cos 3 ω_{2} t + (\frac{3}{4} k_{3} a_{1}^{2} a_{2} + \frac{5}{4} k_{5} a_{1}^{4} a_{2} + \frac{1 5}{8} k_{5} a_{1}^{2} a_{2}^{3}) \cos (ω_{1} \pm 2 ω_{2}) t + (\frac{3}{4} k_{3} a_{1} a_{2}^{2} + \frac{5}{4} k_{5} a_{1} a_{2}^{4} + \frac{1 5}{8} k_{5} a_{1}^{3} a_{2}^{2}) \cos (ω_{2} \pm 2 ω_{1}) t$

4th order components:

$+ \frac{1}{8} k_{4} a_{1}^{4} \cos 4 ω_{1} t + \frac{1}{8} k_{4} a_{2}^{4} \cos 4 ω_{2} t + \frac{1}{2} k_{4} a_{1}^{3} a_{2} \cos (3 ω_{1} \pm ω_{2}) t + \frac{3}{4} k_{4} a_{1}^{2} a_{2}^{2} \cos (2 ω_{1} \pm 2 ω_{2}) t + \frac{1}{2} k_{4} a_{1} a_{2}^{3} \cos (ω_{1} \pm 3 ω_{2}) t$

and 5th order components:

$\begin{matrix} + \frac{1}{1 6} k_{5} a_{1}^{5} \cos 5 ω_{1} t + \frac{1}{1 6} k_{5} a_{2}^{5} \cos 5 ω_{2} t + \frac{5}{1 6} k_{5} a_{1}^{4} a_{2} \cos (4 ω_{1} \pm ω_{2}) t + \frac{5}{8} k_{5} a_{1}^{3} a_{2}^{2} \cos (3 ω_{1} \pm 2 ω_{2}) t + \frac{5}{8} k_{5} a_{1}^{2} a_{2}^{3} \cos (2 ω_{1} \pm 3 ω_{2}) t + \frac{5}{1 6} k_{5} a_{1} a_{2}^{4} \cos (ω_{1} \pm 4 ω_{2}) t \end{matrix}$

The series may be expanded further for terms in k₆e_IN⁶etc. if desired. All the even order terms produce outputs at harmonics of the input signal and that the sum and difference products are well removed in frequency far from the input signal. The odd order products, however, produce signals near the input frequencies ƒ₁±2ƒ₂and ƒ₂±2ƒ₁. Therefore, the odd order intermodulation products cannot be removed by filtering, only by improvement in linearity.

Assuming class A operation, a₁=a₂and k₄, k₅are very small, the 3rd order intermodulation product IM3 becomes proportional to a₃. That means that the cube of the input amplitude and the graph of the intermodulation products will have a slope of 3 in logarithmic scale while the wanted signal will have the slope of 1. Second order products IM2 can be similarly calculated, and the graph for these has a slope of two. The points where these graphs cross are called 3rd order intercept point IP3 and 2nd order intercept point IP2, respectively. IP3 is the point where the intermodulation product is equal to the fundamental signal. This is a purely theoretical consideration, but gives a very convenient method of comparing devices. For example, a device with intermodulation products of −40 dBm at 0 dBm input power is to be compared with one having intermodulation products of −70 dBm for −10 dBm input. By reference to the intercept point, it can be seen that the two devices are equal.

The classical description of intermodulation of analogue radio systems deals with a two-frequency input model to a memoryless non-linear device. This non-linear characteristic can be described by a function ƒ(x), which yields the input-output relation of the element device. The function, ƒ, is usually expanded in a Taylor-series and thus produces the harmonics and as well the linear combinations of the input frequencies. This classical model is well suited to analogue modulation schemes with dedicated frequency lines at the carrier frequencies. The system performance of analogue systems is usually measured in terms of signal-to-noise (S/N) ratio, and the distorting intermodulation signal can adequately be described by a reduction of S/N.

With digital modulation methods, the situation is changed completely. Most digital modulation schemes have a continuous signal spectrum without preferred lines at the carrier frequencies. The system degradation due to intermodulation is measured in terms of bit error ratio (BER) and depends on a variety of system parameters, e.g. the special modulation scheme which is employed. For estimation of the system performance in terms of BER a rigorous analysis of non-linear systems is required. There are two classical methods for the analysis and synthesis of non-linear systems: the first one carries out the expansion of the signal in a Volterra series [27]. The second due to Wiener uses special base functionals for the expansion.

Both methods lead to a description of the non-linear system by higher order transfer functions having n input variables depending on the order of the non-linearity. Two data signals x₁(t) and x₂(t), originated from x(t), are linearly filtered by the devices with the impulse responses h_a(t) and h_b(t) in adjacent frequency bands. The composite summed signal y is hereafter distorted by an imperfect square-law device which might model a transmit-amplifier. The input-output relation of the non-linear device is given by: custom character (t)=y(t)+ay²(t)

The output signal z(t) including the intermodulation noise is caused by non-linearities of third order. For this reason, the imperfect square-law device is now replaced by an imperfect cubic device with the input-output relation: custom character (t)=y(t)+ay³(t)

There are several contributions of the intermodulation noise falling into the used channels near ƒ₀.

Linearization of a transmitter system may be accomplished by a number of methods:

- Feedforward linearization: This technique compares the amplified signal with an appropriately delayed version of the input signal and derives a difference signal, representing the amplifier distortions. This difference signal is in turn amplified, and subtracted from the final HPA output. The main drawback of the method is the requirement for a 2nd amplifier—the technique can, however, deliver an increase in output power of some 3 dB when used with a TWT.
- Feedback linearization: In audio amplifiers, linearization may readily be achieved by the use of feedback, but this is less straightforward at high RF frequencies due to limitations in the available open-loop amplifier gain. It is possible, however, to feedback a demodulated form of the output, to generate adaptive pre-distortion in the modulator. It is clearly not possible to apply such an approach in a bent-pipe transponder, however, where the modulator and HPA are rather widely separated.
- Predistortion: Rather than using a method that responds to the actual instantaneous characteristics of the HPA, it is common to pre-distort the input signal to the amplifier, based on a priori knowledge of the transfer function. Such pre-distortion may be implemented at RF, IF or at baseband. Baseband linearizers, often based on the use of look-up tables held in firmware memory are becoming more common with the ready availability of VLSI techniques, and can offer a compact solution. Until recently, however, it has been easier to generate the appropriate pre-distortion function with RF or IF circuitry.

RF amplifier linearization techniques can be broadly divided into two main categories:

- Open-loop techniques, which have the advantage of being unconditionally stable, but have the drawback of being unable to compensate for changes in the amplifier characteristics.
- Closed-loop techniques, which are inherently self-adapting to changes in the amplifier, but can suffer from stability problems.

Predistortion involves placing a compensating non-linearity into the signal path, ahead of the amplifier to be linearized. The signal is thus predistorted before being applied to the amplifier. If the predistorter has a non-linearity which is the exact inverse of the amplifier non-linearity, then the distortion introduced by the amplifier will exactly cancel the predistortion, leaving a distortionless output. In its simplest analogue implementation, a practical predistorter can be a network of resistors and non-linear elements such as diodes or transistors. Although adaptive predistortion schemes have been reported, where the non-linearity is implemented in digital signal processing (DSP), they tend to be very computationally or memory intensive, and power hungry.

Feedforward [28] is a distortion cancellation technique for power amplifiers. The error signal generated in the power amplifier is obtained by summing the loosely coupled signal and a delayed inverted input signal, so that the input signal component is cancelled. This circuit is called the signal cancelling loop. The error signal is amplified by an auxiliary amplifier, and is then subtracted from the delayed output signal of the power amplifier, so that the distortion at the output is cancelled. This circuit is called the error cancelling loop. It is necessary to attenuate the input signal component lower than the error signal at the input of the auxiliary amplifier, so that the residual main signal does not cause overloading of the auxiliary amplifier, or does not cancel the main signal itself at the equipment output.

Negative feedback [29] is a well-known linearization technique and is widely used in low frequency amplifiers, where stability of the feedback loop is easy to maintain. With multi-stage RF amplifiers however, it is usually only possible to apply a few dB of overall feedback before stability problems become intractable [30]. This is mainly due to the fact that, whereas at low frequency it can be ensured that the open-loop amplifier has a dominant pole in its frequency response (guaranteeing stability), this is not feasible with RF amplifiers because their individual stages generally have similar bandwidths. Of course, local feedback applied to a single RF stage is often used, but since the distortion reduction is equal to the gain reduction, the improvement obtained is necessarily small because there is rarely a large excess of open loop gain available.

At a given center frequency, a signal may be completely defined by its amplitude and phase modulation. Modulation feedback exploits this fact by applying negative feedback to the modulation of the signal, rather than to the signal itself. Since the modulation can be represented by baseband signals, we can successfully apply very large amounts of feedback to these signals without the stability problems that beset direct RF feedback. Early applications of modulation feedback used amplitude (or envelope) feedback only, applied to valve amplifiers [31], where amplitude distortion is the dominant form of non-linearity. With solid-state amplifiers however, phase distortion is highly significant and must be corrected in addition to the amplitude errors.

For estimation of the system performance in terms of BER a rigorous analysis of non-linear systems is required. There are two classical methods for the analysis and synthesis of non-linear systems: the first one carries out the expansion of the signal in a Volterra series [27]. The second due to Wiener uses special base functionals for the expansion. These are the Wiener G-functionals which are orthogonal if white Gaussian noise excites the system. It is the special autocorrelation property of the white Gaussian noise which makes it so attractive for the analysis of non-linear systems. The filtered version of AWGN, the Brownian movement or the Wiener process, has special features of its autocorrelation which are governed by the rules for mean values of the products of jointly normal random variables.

The non-linear system output signal y(t) can be expressed by a Volterra series:

y(t)=H₀+H₁+H₂+ . . .

where H_iis the abbreviated notation of the Volterra operator operating on the input x(t) of the system. The first three operators are given in the following. The convolution integrals are integrated from −∞, to +∞.

H₀[x(t)]=h₀
H₁[x(t)]=∫h₁(τ)x(t−τ)dτ
H₂[x(t)]=∫∫h₂(τ₁,τ₂)x(t−τ₁)x(t−τ₂)dτ₁dτ₂

The kernels of the integral operator can be measured by a variation of the excitation time of input pulses, e.g. for the second order kernel h₂(τ₁, τ₂): x(t)=δ(t−τ₁)δ(t−τ₂). A better method is the measurement of the kernel by the cross-correlation of exciting white Gaussian noise n(t) as input signal with the system output y_i(t). These equations hold, if:

Φ_nn(τ)=Aδ(τ)

is the autocorrelation function of the input signal x(t)=n(t) (white Gaussian noise) where A is the noise power spectral density. The first three kernels are given then by:

$\begin{matrix} h_{0} = \overline{y_{0} (t)} h_{1} (σ) = \frac{1}{A} \overline{y_{1} (t) n (t - σ)} h_{2} (σ_{1}, σ_{2}) = \frac{1}{2 A^{2}} \overline{y_{2} (t) n (t - σ_{1}) n (t - σ_{2})} \end{matrix}$

The overline denotes the expected value, or temporal mean value for ergodic systems.

The method can be expanded to higher order systems by using higher order Volterra operators H_n. However, the Volterra operators of different order are not orthogonal and, therefore, some difficulties arise at the expansion of an unknown system in a Volterra series.

These difficulties are circumvented by the Wiener G-functionals, which are orthogonal to all Volterra operators with lower order, if white Gaussian noise excites the system.

TABLE 1

Volterra

kernels
Direct Fourier Transform
Laplace transform

Linear (1^storder)

H_{1} = ❘ H_{1} (ω) ❘ = ❘ \int_{- \infty}^{+ \infty} h_{1} (τ_{1}) \cdot \exp (- j ω τ_{1}) \cdot d τ_{1} ❘

H₁(p) = k₁· L₁(p)

Quadratic (2^ndorder)

H_{2} = ❘ H_{2} (ω) ❘ = ❘ \int_{- \infty}^{+ \infty} \int_{- \infty}^{+ \infty} h_{2} (τ_{1}, τ_{2}) \cdot \exp [- j ω (τ_{1} + τ_{2})] \cdot d τ_{1} d τ_{2} ❘

H₂(p) = k₂· L₁(2p)

Cubic (3^rdorder)

H_{3} = ❘ H_{3} (ω) ❘ = ❘ \int_{- \infty}^{+ \infty} \int_{- \infty}^{+ \infty} \int_{- \infty}^{+ \infty} h_{3} (τ_{1}, τ_{2}, τ_{3}) \cdot \exp [- j ω (τ_{1} + τ_{2} + τ_{3})] \cdot d τ_{1} d τ_{2} d τ_{3} ❘

H₃(p) = k₃· L₁(3p)

See, Panagiev, Oleg. “Adaptive compensation of the nonlinear distortions in optical transmitters using predistortion.” Radioengineering 17, no. 4 (2008): 55.

The first three Wiener G-functionals are:

G₀[x(t)]=k₀
G₁[x(t)]=∫k₁(τ₁)x(t−τ₁)dτ₁
G₂[x(t)]=∫∫k₂(τ₁,τ₂)x(t−τ₁)x(t−τ₂)dτ₁dτ₂−A∫k₂(τ₁,τ₂)dτ₁
G₃[x(t)]=∫∫∫k₃(τ₁,τ₂,τ₃)x(t−τ₁)x(t−τ₂)x(t−τ₃)dτ₁dτ₂dτ₃−3A∫∫k₃(τ₁,τ₂,τ₂)x(t−τ₁)dτ₁dτ₂

For these functionals hold:

H_m[n(t)]G_n[n(t)]=0 for m<n

if the input signal n(t) is white Gaussian noise.

The two data signals x₁(t) and x₂(t), from a single signal x(t), are linearly filtered by the devices with the impulse responses h_a(t) and h_b(t) in adjacent frequency bands. The composite summed signal y is hereafter distorted by an imperfect square-law device which might model a transmit-amplifier. The input-output relation of the non-linear device is given by:

custom character (t)=y(t)+ay²(t)

The output signal z(t) is therefore determined by:

custom character (t)=∫[h_a(τ)+h_b(τ)]x(t−τ)dτ+a{∫[h_a(τ)+h_b(τ)]x(t−τ)dτ}²

The first and second order Volterra-operators H₁and H₂for this example are accordingly determined by the kernels:

h₁(τ)=h_a(τ)+h_b(τ)

and

h₂(τ₁,τ₂)=h_a(τ₁)[h_a(τ₂)+h_b(τ₂)]+h_b(τ₁)[h_a(τ₂)+h_b(τ₂)]

This kernel h₂(τ₁, τ₂) is symmetric, so that:

h₂(τ₁,τ₂)=h₂(τ₂,τ₁)

The second order kernel transform H₂(ω₁, ω₂) is obtained by the two-dimensional Fourier-transform with respect to τ₁and τ₂, and can be obtained as:

H₂(ω₁,ω₂)={H_a(ω₁)[H_a(ω₂)+H_b(ω₂)]+H_b(ω₁)[H_a(ω₂)+H_b(ω₂)]}

by elementary manipulations. H_a(ω) and H_b(ω) are the Fourier-transforms of h_a(t) and h_b(t). With the transform X(ω) of the input signal x(t), an artificial two-dimensional transform Z₂(ω₁, ω₂) is obtained:

Z₍₂₎(ω₁,ω₂)=H₂(ω₁,ω₂)X(ω₁)X(ω₂)

with the two-dimensional inverse Z₂(t₁, t₂). The output signal z(t) is:

z(t)=z₍₂₎(t,t)

The transform Z(ω) of z(t) can be obtained by convolution:

$\begin{matrix} Z (ω) = \frac{1}{2 π} \int Z_{(2)} (ω_{1}, ω - ω_{1}) d ω_{1} \end{matrix}$

where the integration is carried out from −∞ to +∞.

The output z(t) can be as well represented by use of the Wiener G-functionals:

z(t)=G₀+G₁+G₂+ . . .

where G_iis the simplified notation of G_i[x(t)]. The first two operators are:

G₀[x(t)]=−A∫[h_a(τ)+h_b(τ)]²dτ=const
G₁[x(t)]=∫[h_a(τ)+h_b(τ)]x(t−τ)dτ

The operator G_iequals H_iin this example. For x(t) equal white Gaussian noise x(t)=n(t):

G₁[n(t)]h₀holds for all h₀, especially:

G₁G₀=0.
G₂[x(t)]=∫[h_a(τ₁)h_a(τ₂)+h_a(τ₁)h_b(τ₂)+h_b(τ₁)h_a(τ₂)+h_b(τ₁)h_b(τ₂)]x(t−τ₁)x(t−τ₂)dτ₁dτ₂−A∫[h_a(τ₁)+h_b(τ₁)]²dτ₁

The consequence is:

G₂h₀=h₀∫[h_a(τ₁)h_a(τ₂)+h_a(τ₁)h_b(τ₂)+h_b(τ₁)h_a(τ₂)+h_b(τ₁)h_b(τ₂)]n(t−τ₁)n(t−τ₂)dτ₁dτ₂−h₀A∫[h_a(τ₁)+h_b(τ₁)]²dτ₁

and

G₂h₀=0 because of n(t−τ₁)n(t−τ₂)=Aδ(τ₁−τ₂)

and similarly:

G₂H₁=0 for all H₁

This equation involves the mean of the product of three zero mean jointly Gaussian random variables, which is zero.

The Wiener kernels can be determined by exciting the system with white Gaussian noise and taking the average of some products of the system output and the exciting noise process n(t):

$k_{0} = \overline{z (t)} k_{1} (τ) = \frac{1}{A} \overline{z (t) n (t - τ)} and k_{2} (τ_{1}, τ_{2}) = \frac{1}{2 A^{2}} \overline{z (t) n (t - τ_{1}) n (t - τ_{2})}$

For RF-modulated signals the intermodulation distortion in the proper frequency band is caused by non-linearities of third order. For this reason, the imperfect square-law device is now replaced by an imperfect cubic device with the input-output relation:

custom character (t)=y(t)+ay³(t)

If only the intermodulation term which distorts the signal in its own frequency band is considered, the kernel transform of the third-order Volterra operator Z₍₃₎(ω₁, ω₂, ω₃) becomes then:

$\begin{matrix} Z_{(3)} (ω_{1}, ω_{2}, ω_{3}) = a \prod_{i = 1}^{3} [H_{a} (ω_{i}) + H_{b} (ω_{i})] X (ω_{i}) \end{matrix}$

The intermodulation part in the spectrum of z(t) is now given by:

$\begin{matrix} Z (ω) = \frac{1}{{(2 π)}^{2}} \int \int Z_{(3)} (ω - μ_{1}, μ_{1} - μ_{2}, μ_{2}) d μ_{1} d μ_{2} \end{matrix}$

For a cubic device replacing the squarer, however, there are several contributions of the intermodulation noise falling into the used channels near ƒ₀.

See, Amplifier References, infra.

The Volterra series is a general technique, and subject to different expressions of analysis, application, and simplifying presumptions. Below is further discussion of the technique.

A system may have hidden states of input-state-output models. The state and output equations of any analytic dynamical system are

{dot over (x)}(t)=ƒ(x,u,θ)
y(t)=g(x,u,θ)+ε

{dot over (x)}(t) is an ordinary differential equation and expresses the rate of change of the states as a parameterized function of the states and input. Typically, the inputs u(t) correspond to designed experimental effects. There is a fundamental and causal relationship (Fliess et al 1983) between the outputs and the history of the inputs. This relationship conforms to a Volterra series, which expresses the output y(t) as a generalized convolution of the input u(t), critically without reference to the hidden states {dot over (x)}(t). This series is simply a functional Taylor expansion of the outputs with respect to the inputs (Bendat 1990). The reason it is a functional expansion is that the inputs are a function of time.

$y (t) = \sum_{i} \int_{0}^{t} \dots \int_{0}^{t} κ_{i} (σ_{1}, \dots, σ_{i}) u (t - σ_{1}), \dots, u (t - σ_{i}) d σ_{1}, \dots, d σ_{i} κ_{i} (σ_{1}, \dots, σ_{i}) = \frac{\partial^{i} y (t)}{\partial u (t - σ_{1}), \dots, \partial u (t - σ_{i})}$

where κ_i(σ₁, . . . σ_i) is the ith order kernel, and the integrals are restricted to the past (i.e., integrals starting at zero), rendering the equation causal. This equation is simply a convolution and can be expressed as a GLM. This means that we can take a realistic model of responses and use it as an observation model to estimate parameters using observed data. Here the model is parameterized in terms of kernels that have a direct analytic relation to the original parameters θ of the physical system. The first-order kernel is simply the conventional HRF. High-order kernels correspond to high-order HRFs and can be estimated using basis functions as described above. In fact, by choosing basis function according to

${A (σ)}_{i} = \frac{\partial {κ (σ)}_{1}}{\partial θ_{i}}$

one can estimate the physical parameters because, to a first order approximation, β_i=θ_i. The critical step is to start with a causal dynamic model of how responses are generated and construct a general linear observation model that allows estimation and inference about the parameters of that model. This is in contrast to the conventional use of the GLM with design matrices that are not informed by a forward model of how data are caused.

Dynamic causal models assume the responses are driven by designed changes in inputs. An important conceptual aspect of dynamic causal models pertains to how the experimental inputs enter the model and cause responses. Experimental variables can illicit responses in one of two ways. First, they can elicit responses through direct influences on elements. The second class of input exerts its effect through a modulation of the coupling among elements. These sorts of experimental variables would normally be more enduring. These distinctions are seen most clearly in relation to particular forms of causal models used for estimation, for example the bilinear approximation

$\begin{matrix} \begin{matrix} \dot{x} (t) = f (x, u) \\ = Ax + u B x + Cu \end{matrix} y = g (x) + ε A = \frac{\partial f}{\partial x} B = \frac{\partial^{2} f}{\partial x \partial u} C = \frac{\partial f}{\partial u} \end{matrix}$

This is an approximation to any model of how changes in one element x(t)_iare caused by activity of other elements. Here the output function g(x) embodies a model. The matrix A represents the connectivity among the regions in the absence of input u(t). Effective connectivity is the influence that one system exerts over another in terms of inducing a response ∂{dot over (x)}/∂x. This latent connectivity can be thought of as the intrinsic coupling in the absence of experimental perturbations. The matrix B is effectively the change in latent coupling induced by the input. It encodes the input-sensitive changes in A or, equivalently, the modulation of effective connectivity by experimental manipulations. Because B is a second-order derivative it is referred to as bilinear. Finally, the matrix C embodies the extrinsic influences of inputs on activity. The parameters θ={A,B,C} are the connectivity or coupling matrices that we wish to identify and define the functional architecture and interactions among elements. We can express this as a GLM and estimate the parameters using EM in the usual way (see Friston et al 2003). Generally, estimation in the context of highly parameterized models like DCMs requires constraints in the form of priors. These priors enable conditional inference about the connectivity estimates.

The central idea, behind dynamic causal modelling (DCM), is to model a physical system as a deterministic nonlinear dynamic system that is subject to inputs and produces outputs. Effective connectivity is parameterized in terms of coupling among unobserved states. The objective is to estimate these parameters by perturbing the system and measuring the response. In these models, there is no designed perturbation and the inputs are treated as unknown and stochastic. Furthermore, the inputs are often assumed to express themselves instantaneously such that, at the point of observation the change in states will be zero. In the absence of bilinear effects we have

$\begin{matrix} \dot{x} = 0 \\ = Ax + Cu \end{matrix} x = - A^{- 1} C u$

This is the regression equation used in SEM where A=A′−I and A′ contains the off-diagonal connections among regions. The key point here is that A is estimated by assuming u is some random innovation with known covariance. This is not really tenable for designed experiments when u represent carefully structured experimental inputs. Although SEM and related autoregressive techniques are useful for establishing dependence among responses, they are not surrogates for informed causal models based on the underlying dynamics of these responses.

The Fourier transform pair relates the spectral and temporal domains. We use the same symbol F, although F(t) and F(ω) are different functions:

$F (t) = \frac{1}{2 π} \int_{- \infty}^{\infty} d ω F (ω) e^{- i ω t}, F (ω) = \int_{- \infty}^{\infty} d t F (t) e^{i ω t}$

Accordingly, a convolution integral is derived:

$D (t) = \int_{- \infty}^{\infty} {dt}_{1} ε (t_{1}) E (t - t_{1})$

where D(t), ε(t), E(t), are related to D(ω), ε(−iω), E(ω), respectively. Note that D(t) can be viewed as an integral operation, acting on E(t) is the simplest form of a Volterra Function Series (VFS). This can also be expressed in the VDO representation

D(t)=ε(∂_t)E(t)=ε(∂_τ)E(τ)|_τ⇒t

The instruction τ⇒t is superfluous in a linear case, but becomes important for non-linear systems. For example, consider a harmonic signal clarifying the role of the VDO:

$E (t) = E_{0} e^{- i ω t} D (t) = E_{0} e^{- i ω t} \int_{- \infty}^{\infty} d t_{1} ε (t_{1}) e^{i ω t_{1}} = ε (- i ω) E_{0} e^{- i ω t} = ε (\partial_{t}) E_{0} e^{- i ω t}$

In nonlinear systems, the material relations involve powers and products of fields, and {dot over (x)}(t) can be replaced by a series involving powers of E(ω), but this leads to inconsistencies.

However, the convolution can be replaced by a “super convolution”, the Volterra function series (VFS), which can be considered a Taylor expansion series with memory, given by:

$D (t) = \sum_{m} D^{(m)} (t) D^{(m)} (t) = \int_{- \infty}^{\infty} {dt}_{1} \dots \int_{- \infty}^{\infty} {dt}_{m} ε^{(m)} (t_{1}, \dots, t_{m}) E (t - t_{1}) \dots E (t - t_{m})$

Typically, the VFS contains the products of fields expected for nonlinear systems, combined with the convolution structure. Various orders of nonlinear interaction are indicated by m. Theoretically all the orders co-exist (in practice the series will have to be truncated within some approximation), and therefore we cannot readily inject a time harmonic signal. If instead a periodic signal,

$E (t) = \sum_{n} E_{n} e^{- i n ω t}$

is provided, we find

$D^{(m)} (t) = \sum_{n_{1}, \dots, n_{m}} ε^{(m)} (- i n_{1} ω, \dots, - i n_{m} ω) E_{n_{1}} \dots E_{n_{m}} e^{- iN ω t} = \sum_{N} D_{N} e^{- iN ω t} N = n_{1} + \dots + n_{m}$

displaying the essential features of a nonlinear system, namely, the dependence on a product of amplitudes, and the creation of new frequencies as sums (including differences and harmonic multiples) of the interacting signals frequencies. This function contains the weighting function ε^(m)(−in₁ω, . . . , −in_mω) for each interaction mode.

The extension to the nonlinear VDO is given by

D^(m)(t)=ε^(m)(∂_t₁, . . . ,∂_t_m)E(t₁) . . . E(t_m)|_t₁_{, . . . ,t}_m_⇒t

In which the instruction t₁, . . . , t_m⇒t guarantees the separation of the differential operators, and finally renders both sides of the equation to become functions of t.

The VFS, including the convolution integral, is a global expression describing D(t) as affected by integration times extending from −∞ to ∞. Physically this raises questions about causality, i.e., how can future times affect past events. In the full-fledged four-dimensional generalization causality is associated with the so called “light cone” (Bohm, 1965). It is noted that the VDO representation is local, with the various time variables just serving for book keeping of the operators, and where this representation is justified, causality problems are not invoked. In a power amplifier the physical correlate of this feature is that all past activity leads to a present state of the system, e.g., temperature, while the current inputs affect future states. In general, the frequency constraint is obtained from the Fourier transform of the VFS, having the form

$\begin{matrix} D^{(m)} (ω) = \frac{1}{{(2 π)}^{m - 1}} \int_{- \infty}^{\infty} d ω_{1} \dots \int_{- \infty}^{\infty} d ω_{m - 1} ε^{(m)} (- i ω_{1}, \dots, - i ω_{m}) E (ω_{1}) \dots E (ω_{m}) ω = ω_{1} + \dots + ω_{m} \end{matrix}$

In which we have m−1 integrations, one less than in the VFS form. Consequently, the left and right sides of the Fourier transform are functions of ω, ω_m, respectively. The additional constraint ω=ω₁+ . . . +ω_mcompletes the equation and renders it self-consistent.

See, Volterra Series References, infra.

An alternate analysis of the VFS is as follows. Let x[n] and y[n] represent the input and output signals, respectively, of a discrete-time and causal nonlinear system. The Volterra series expansion for y[n] using x[n] is given by:

$y [n] = h_{0} + \sum_{m_{1} = 0}^{\infty} h_{1} [m_{1}] x [n - m_{1}] + \sum_{m_{1} = 0}^{\infty} \sum_{m_{2} = 0}^{\infty} h_{2} [m_{1}, m_{2}] x [n - m_{1}] x [n - m_{2}] + \dots + \sum_{m_{1} = 0}^{\infty} \sum_{m_{2} = 0}^{\infty} \dots \underset{m_{p} = 0}{\sum^{\infty}} h_{p} [m_{1}, m_{2}, \dots, m_{p}] x [n - m_{1}] \times [n - m_{2}] \dots x [n - m_{p}] + \dots)$

h_p[m₁, m₂, . . . , m_p] is known as the p-th order Volterra kernel of the system. Without any loss of generality, one can assume that the Volterra kernels are symmetric, i.e., h_p[m₁, m₂, . . . , m_p] is left unchanged for any of the possible p! Permutations of the indices m₁, m₂, . . . , m_p. One can think of the Volterra series expansion as a Taylor series expansion with memory. The limitations of the Volterra series expansion are similar to those of the Taylor series expansion, and both expansions do not do well when there are discontinuities in the system description. Volterra series expansion exists for systems involving such type of nonlinearity. Even though clearly not applicable in all situations, Volterra system models have been successfully employed in a wide variety of applications.

Among the early works on nonlinear system analysis is a very important contribution by Wiener. His analysis technique involved white Gaussian input signals and used “G-functionals” to characterize nonlinear system behavior. Following his work, several researchers employed Volterra series expansion and related representations for estimation and time-invariant or time variant nonlinear system identification. Since an infinite series expansion is not useful in filtering applications, one must work with truncated Volterra series expansions.

The discrete time impulse response of a first order (linear) system with memory span is aggregate of all the N most recent inputs and their nonlinear combinations into one expanded input vector as

X_e(n)=[x(n),x(n−1), . . . ,x(n−N+1),x²(n)x(n)x(n−1), . . . ,x^Q(n−N+1)]^T

Similarly, the expanded filter coefficients vector H(n) is given by

H(n)=[h₁(0),h₁(1), . . . ,h₁(N−1),h₂(0,0),h₂(0,1), . . . ,h_Q(N−1, . . . ,N−1)]^T

The Volterra Filter input and output can be compactly rewritten as

y(n)=H^T(n)X_e(n)

The error signal e(n) is formed by subtracting y(n) from the noisy desired response d(n), i.e.,

e(n)=d(n)−y(n)=d(n)−H^T(n)X_e(n)

For the LMS algorithm, this may be minimized to

E[e²(n)]=E[d(n)−H^T(n)X_e(n)]

The LMS update equation for a first order filter is

H(n+1)=H(n)+μ|e(n)|X_e(n)

where μ is small positive constant (referred to as the step size) that determines the speed of convergence and also affects the final error of the filter output. The extension of the LMS algorithm to higher order (nonlinear) Volterra filters involves a few simple changes. Firstly, the vector of the impulse response coefficients becomes the vector of Volterra kernels coefficients. Also, the input vector, which for the linear case contained only a linear combination, for nonlinear time varying Volterra filters, complicates treatment.

The RLS (recursive least squares) algorithm is another algorithm for determining the coefficients of an adaptive filter. In contrast to the LMS algorithm, the RLS algorithm uses information from all past input samples (and not only from the current tap-input samples) to estimate the (inverse of the) autocorrelation matrix of the input vector.

To decrease the influence of input samples from the far past, a weighting factor for the influence of each sample is used. The Volterra filter of a fixed order and a fixed memory adapts to the unknown nonlinear system using one of the various adaptive algorithms. The use of adaptive techniques for Volterra kernel estimation has been well studied. Most of the previous research considers 2nd order Volterra filters and some consider the 3rd order case.

A simple and commonly used algorithm is based on the LMS adaptation criterion. Adaptive Volterra filters based on the LMS adaptation algorithm are computational simple but suffer from slow and input signal dependent convergence behavior and hence are not useful in many applications. As in the linear case, the adaptive nonlinear system minimizes the following cost function at each time:

$J [n] = \sum_{k = 0}^{n} {λ^{n - k} (d [k] - H^{T} [n] X [k])}^{2}$

where, H(n) and X(n) are the coefficients and the input signal vectors, respectively, A is a factor that controls the memory span of the adaptive filter and d(k) represents the desired output. The solution can be obtained by differentiating J[n] with respect to H[n], setting the derivative to zero, and solving for H[n]. The optimal solution at time n is given by

H[n]=C⁻¹[n]P[n]

where,

$C [n] = \sum_{k = 0}^{n} λ^{n - k} X [k] X^{T} [k] and P [n] = \sum_{k = 0}^{n} λ^{n - k} d [k] X [k]$

H[n] can be recursively updated by realizing that

C[n]=λC[n−1]+X[n]X^T[n] and P[n]=λP[n−1]+d[n]X[n]

The computational complexity may be simplified by making use of the matrix inversion lemma for inverting C[n]. The derivation is similar to that for the RLS linear adaptive filter.

C⁻¹[n]=λ⁻¹C⁻¹[n−1]−λ⁻¹k[n]X^T[n]C⁻¹[n−1]

There are a few simple models for basic amplifier non-linear behavior. A more rigorous model could include the Volterra series expansion which can model complex non-linearities such as memory effects. Among the simpler models are the Rapp model, Saleh model and the Ghorbani model. Combinations of pure polynomial models and filter models are also often referred to as fairly simple models, e.g., the Hammerstein model.

The advantage of the simpler models is usually in connection to for a need of very few parameters to model the non-linear behavior. The drawback is that such a model only can be used in conjunction with simple architecture amplifiers such as the basic Class A, AB and C amplifiers. Amplifiers such as the high efficiency Doherty amplifier can in general not be modelled by one of these simple models. In addition, to properly capture the PA behavior for the envisaged large NR bandwidths, it is essential to use PA models capturing the memory effects. Such models would require an extensive set of empirical measurements for proper parameterization.

The Rapp model has basically two parameters by which the general envelop distortion may be described. It mimics the general saturation behavior of an amplifier and lets the designer set a smoothness of the transition by a P-factor. By extending this also to model phase distortion, one has in total six parameters available. The basic simple model may be found as:

$V_{out} = \frac{V_{i n}}{{(1 + {(\frac{❘ V_{i n} ❘}{V_{sat}})}^{2 P})}^{\frac{1}{2 P}}}$

This model produces a smooth transition for the envelope characteristic as the input amplitude approaches saturation. In the more general model, both AM-AM and AM-PM distortion can be modelled. In general terms, the model describes the saturation behavior of a radio amplifier in a good way.

$F_{A M - A M} = \frac{G x}{{(1 + {❘ \frac{G x}{V_{sat}} ❘}^{2 P})}^{\frac{1}{2 P}}} F_{A M - P M} = \frac{A x^{q}}{1 + {(\frac{x}{B})}^{q}}$

where “x” is the envelope of the complex input signal. If signal measurements are at hand of the input/output relationship, the parameters of the model may be readily found for a particular amplifier by for example regression techniques. The strength of the Rapp model is lies in its simple and compact formulation, and that it gives an estimation of the saturation characteristics of an amplifier. The drawback of this simple model is of course that it cannot model higher order classes of amplifiers such as the Doherty amplifier. It also lacks the ability to model memory effects of an amplifier.

The Saleh model is a similar model to the Rapp model. It also gives an approximation to the AM-AM and AM-PM characteristics of an amplifier. It offers a slightly fewer number of parameters (4) that one can use to mimic the input/output relationship of the amplifier. The AM-AM distortion relation and AM-PM distortion relation are found to be as:

${g (r)}_{A M - A M} = \frac{α_{a} r}{1 + β_{a} r^{2}} {f (r)}_{A M - P M} = \frac{α_{φ} r^{2}}{1 + β_{φ} r^{2}}$

where “r” is the envelope of the complex signal fed into the amplifier, and α/β are real-valued parameters that can be used to tune the model to fit a particular amplifier.

The Ghorbani model also gives expressions similar to the Saleh model, where AM-AM and AM-PM distortion is modeled. Following Ghorbani, the expressions are symmetrically presented:

$g (r) = \frac{x_{1} r^{x_{2}}}{1 + x_{3} r^{x_{2}}} + x_{4} r$

$f (r) = \frac{y_{1} r^{y_{2}}}{1 + y_{3} r^{y_{2}}} + y_{4} r$

In the expressions above, g(r) corresponds to AM-AM distortion, while ƒ(r) corresponds to AM-PM distortion. The actual scalars x_1-4and y_1-4have to be extracted from measurements by curve fitting or some sort of regression analysis.

The next step in the more complex description of the non-linear behavior of an amplifier is to view the characterization as being subject to a simple polynomial expansion. This model has the advantage that it is mathematically pleasing in that it for each coefficient reflects higher order of inter-modulations. Not only can it model third order intermodulation, but also fifth/seventh/ninth etc. Mathematically it can also model the even order intermodulation products as well, it merely is a matter of discussion whether these actually occur in a real RF application or not.

$y (t) = a_{0} + a_{1} x (t) + a_{2} {x (t)}^{2} + a_{3} {x (t)}^{3} + a_{4} {x (t)}^{4} \dots \dots$

$A_{IP 3} = \sqrt{4 a_{1} / 3 ❘ a_{3} ❘}$

Coefficients may be readily expressed in terms of Third Order Intercept point IP3 and gain, as described above. This feature makes this model especially suitable in low level signal simulations, since it relates to quantities that usually are readily available and easily understood amongst RF engineers.

The Hammerstein model consists of a combination of a Linear+Non-Linear block that is capable of mimicking a limited set of a Volterra Series. As the general Volterra series models a nested series of memory and polynomial representations, the Hammerstein model separates these two defining blocks that can in theory be separately identified with limited effort. The linear part is often modelled as a linear filter in the form of a FIR-filter.

$s (n) = \sum_{k = 0}^{K - 1} h (k) x (n - k)$

The non-linear part is then on the other hand simply modelled as polynomial in the enveloped domain.

y(t)=a₀+a₁x(t)+a₂x(t)²+a₃x(t)³+a₄x(t)⁴. . .

The advantage of using a Hammerstein model in favor of the simpler models like Rapp/Saleh or Ghorbani is that it can in a fairly simple way also model memory effects to a certain degree. Although, the model does not benefit from a clear relationship to for example IIP3/Gain but one has to employ some sort of regression technique to derive polynomial coefficients and FIR filter tap coefficients.

The Wiener model describes like the Hammerstein model a combination of Non-linear and Linear parts that are cascaded after each other. The difference to the Hammerstein model lies in the reverse order of non-linear to linear blocks. The initial non-linear block is preferably modelled as a polynomial in the envelope of the complex input signal. This block is the last one in the Hammerstein model as described above. The polynomial coefficients may themselves be complex, depending on what fits measured data best. See expressions for non-linear and linear parts under the Hammerstein section. The second block which is linear may be modelled as an FIR filter with a number of taps that describes the memory depth of the amplifier.

The state-of-the-art approaches consider the so called Volterra series, and is able to model all weak non-linearity with fading memory. Common models like, for example, the memory polynomial can also be seen as a subset of the full Volterra series and can be very flexible in designing the model by simply adding or subtracting kernels from the full series.

The discrete-time Volterra series, limited to causal systems with symmetrical kernels (which is most commonly used for power amplifier modelling) is written as

$y [n] = β_{0} + \sum p = 1 \sum τ_{1} = 0 \sum τ_{2} = τ_{1} \dots \sum τ_{p} = τ_{p - 1} β_{p, τ_{1}, τ_{2}, \dots, τ_{p}} \prod_{j_{1} = 1}^{p} x [n - τ_{j_{1}} \prod_{j_{2} = p + 1}^{2 p - 1} \overline{x} [n - τ_{j_{2}}]$

in which P is the non-linear order and M is the memory-depth. There are benefits which the Volterra series hold over other modelling approaches, including:

- It is linear in parameters, meaning that the optimal parameters may be found through simple linear regression analysis from measured data. It further captures frequency dependencies through the inclusion of memory effects which is a necessity for wideband communication.
- The set of kernels, or basis functions, best suited for modelling a particular power amplifier may be selected using methods which rely on physical insight. This makes the model scalable for any device technology and amplifier operation class.
- It can be extended into a multivariate series expansion in order to include the effects of mutual coupling through antenna arrays. This enables the studies on more advanced algorithms for distortion mitigation and pre-coding.

It may be observed that other models such as static polynomials, memory polynomials and combinations of the Wiener and Hammerstein models are all subsets of the full Volterra description. As previously stated, empirical measurements are needed to parameterized PA model based on Volterra series expansion.

A subset of the Volterra Series is the memory polynomial with polynomial representations in several delay levels. This is a simpler form of the general Volterra series. The advantage of this amplifier model is its simple form still taking account of memory effects. The disadvantage is that the parameters have to be empirically solved for the specific amplifier in use.

PA_memory=x(t)·[a₀+a₁·|x(t)|+a₂·|x(t)|²+ . . . ]++x(t−t₀)·[b₀+b₁·|x(t−t₀)|+b₂·|x(t−t₀)|²+ . . . ]++x(t−t₁)·[c₀+c₁·|x(t−t₁)|+c₂·|x(t−t₁)|²+ . . . ]+ . . .

The equation above shows an expression for a memory polynomial representation of an amplifier involving two memory depth layers. Each delayed version of the signal is associated with its own polynomial expressing the non-linear behavior.

See Filter References, infra.

The purpose of a PA behavioral model is to describe the input-to-output relationship as accurately as possible. State-of-the-art approaches lean on a fundament of the so called Volterra series consisting of a sum of multidimensional convolutions. Volterra series are able to model all weak nonlinearities with fading memory and thus are feasible to model conventional PAs aimed for linear modulation schemes.

The GMP model is given by

$y_{GMP} (n) = \sum_{k \in K_{a}} \sum_{l \in L_{a}} a_{kl} x (n - l) {❘ x (n - l) ❘}^{2 k} + \sum_{k \in K_{b}} \sum_{l \in L_{b}} \sum_{m \in M} b_{klm} x (n - l) {❘ x (n - l - m) ❘}^{2 k}$

where y_GMP(n) and x(n) represent the complex baseband equivalent output and input, respectively, of the model. The first term represents the double sum of so-called diagonal terms where the input signal at time shift l, x(n−l); l∈L_a, is multiplied by different orders of the time aligned input signal envelope |x(n−l)|^2k; k∈K_a. The triple sum represents cross terms, i.e. the input signal at each time shifts is multiplied by different orders of the input signal envelope at different time shifts. The GMP is linear in the coefficients, a_kland b_klm, which provides robust estimation based on input and output signal waveforms of the PAs to be characterized. As a complement to the above, also memoryless polynomial models have been derived based on:

$y_{P} (n) = \sum_{k \in K_{p}} a_{k} x (n) {❘ x (n) ❘}^{2 k}$

It is thus seen that, while the Volterra series has been considered generally in a variety of contexts, and for power amplifier linearization, the particular implementation does not necessarily follow from broad prescriptions.

See, Volterra Series Patents, infra.

SUMMARY OF THE INVENTION

A deep neural network (DNN)-based equalizer is provided to equalize the PA distorted signals at a radio frequency receiver. This DNN equalizer exploits both the Volterra series nonlinearity modeling of PAs, to construct the input features of the DNN, which can help the DNN converge rapidly to the desired nonlinear response under limited training data and training.

Conventionally, Volterra series and neural networks are studied as two separate techniques for nonlinear PAs [2]. Volterra series has been a popular choice for constructing the models of nonlinear power amplifiers. Many digital predistorters or nonlinear equalizers have been developed based on such modeling. Similarly, artificial neural networks have also been applied to model or equalizer the nonlinear PAs. By integrating these two techniques together, equalizers may be more efficient and have low-cost implementation than conventional digital pre-distorters, and have high performance in mitigating power amplifier with even severe nonlinearity.

In particular, conventional shallow feedforward neural networks using time-delayed inputs have only limited performance. The present DNN equalizer has much superior performance and does not need too much training data.

Nonlinear Power Amplifier Models

The nonlinear response of the power amplifiers are usually described by the baseband discrete model y(n)=ƒ(x(n)), where x(n) is the input signal and y(n) is the output signal. The function ƒ(x(n)) is some nonlinear function.

Consider the baseband discrete model of the PA y(n)=ƒ(x(n), x(n−1), . . . ), where x(n) is the input signal, y(n) is the output signal, and ƒ(⋅) is some nonlinear function. The simplest nonlinear PA model is the “AM-AM AM-PM” model. Let the amplitude of the input signal be Vx=E[|x(n)|], where E[⋅] denotes short-term expectation or average. The output sample y(n)'s amplitude V_y=E[y(n)] and additional phase change ψ_y=E[/y(n)] depend on V_xin nonlinear ways as

$\begin{matrix} V_{y} = \frac{g V_{x}}{{(1 + \frac{g V_{x}}{c})}^{\frac{1}{2 σ}}}, V_{y} = \frac{α V_{x}^{p}}{{(1 + \frac{V_{x}}{β})}^{q}} & (1) \end{matrix}$

where g is the linear gain, σ the smoothness factor, and c denotes the saturation magnitude of the PA. Typical examples of these parameters are g=4:65, σ=0:81, c=0:58, α=2560, β=0:114, p=2:4, and q=2:3, which are used in the PA models regulated by IEEE 803.11ad task group (TG) [10].

More accurate models should take into consideration the fact that nonlinearity leads to memory effects. In this case, Volterra series are typically used to model PAs [4] [21]. A general model is [5]

$\begin{matrix} y (n) = \sum_{d = 0}^{D} \sum_{k = 0}^{P} b_{k d} x (n - d) {❘ x (n - d) ❘}^{k - 1} & (2) \end{matrix}$

with up to P^thorder nonlinearity and up to D step memory.

Because higher order nonlinearity usually has smaller magnitudes, in order to simplify models, many papers have considered smaller P only, e.g.,

$y (n) = \sum_{d = 0}^{D} (β_{D} x (n - d) + α_{d} x (n - d) {❘ x (n - d) ❘}^{2})$

with only the third-order nonlinearity. It can be shown that only odd-order nonlinearity (i.e., odd k) is necessary as even-order nonlinearity disappears during spectrum analysis.

It can be shown that only odd-order nonlinearity (i.e., odd k) is necessary because even-order nonlinearity falls outside of the passband and will be filtered out by the receiver bandpass filters [2]. To illustrate this phenomenon, we can consider some simple examples where the input signal x(n) consists of a few single frequency components only. Omitting the memory effects, if x(n) is a single frequency signal, i.e., x(n)=V₀cos(a₀+ϕ), where a₀=2πƒ₀n. Then, the output signal can be written as

$\begin{matrix} y (n) = c_{1} V_{0} \cos (a_{0} + ϕ + ψ_{1}) + (\frac{3}{4} c_{3} V_{0}^{3} + \frac{5}{8} c_{5} V_{0}^{5}) \cos (a_{0} + ϕ + ψ_{3} + ψ_{5}) & (3) \end{matrix}$

$\begin{matrix} + \frac{1}{2} c_{2} V_{0}^{2} + \frac{3}{8} c_{4} V_{0}^{4} & (4) \end{matrix}$

$\begin{matrix} + (\frac{1}{2} c_{2} V_{0}^{2} + \frac{1}{2} c_{4} V_{0}^{4} (\cos (2 a_{0} + 2 ϕ + 2 ψ_{2} + 2 ψ_{4}) \dots & (5) \end{matrix}$

where the first line (3) is the inband response with AM-AM & AM-PM nonlinear effects, the second line (4) is the DC bias, and the third line (5) includes all the higher frequency harmonics. At the receiving side, we may just have (3) left because all the other items will be canceled by bandpass filtering.

If x(n) consists of two frequencies, i.e., x(n)=V₁cos(a₁+ϕ₁)+V2 cos(a₂+ϕ₂), where a_i=2πƒ_in, then the inband response includes many more items, such as the first order items c_iV_icos(a_i+ϕ_i+ψ_i), the third order items c₃(V_i³+V_iV_j²) cos(aⁱ+ϕ_i+ψ_i), the fifth order items c₅(V_i⁵+V_iV_j⁴+V_iV_j²) cos(a_i+ϕ_i+ψ_i), for i,j∈{1,2}. There are also intermodulation items that consist of na_i±ma_jas long as they are within the passband of the bandpass filter, such as (V_i²V_j+V_i²V_j³+V₁⁴V_j) cos(2a_i−a_j+2ϕ_i−ϕ_j+2ψ_i−ψ_j).

There are many other higher order items with frequencies na_i, n(a_i±a_j), or na_i+ma_j, that cannot pass the passband filter. One of the important observations is that the contents that can pass the passband filter consist of odd-order nonlinearity only.

If x(n) consists of three or more frequencies, we can have similar observations, albeit the expressions are more complex. Let the input signal x(n) be

$\begin{matrix} x (n) = \sum_{i = 1}^{3} V_{i} \cos (a_{i}), a_{i} = 2 π f_{i} n & (6) \end{matrix}$

Based on [22], the nonlinear distorted output response y(n)=ƒ(x(n)) can be written as

$\begin{matrix} y (n) = \sum_{i = 1}^{\infty} k_{i} x_{i} (n) & (7) \end{matrix}$

where k_irepresents the gain coefficients for the i^thorder components. The 1st order component is simply k₁x(n). The 2nd order component includes the DC component, the sum/difference of beat components, and the second-order harmonic components. Specifically,

k₂x²(n)=g_2,0+g_2,1(n)+g_2,2(n), (8)

where

$g_{2, 0} = \sum_{i = 1}^{3} V_{i}^{2} / 2 (n)$

$g_{2, 1} = \sum_{i = 1}^{3} \sum_{j \neq i} V_{i} V_{j} \cos (a_{i} \pm a_{j})$

$g_{2, 2} = \sum_{i = 1}^{3} V_{i}^{2} \cos (2 a_{i}) / 2.$

The 3rd order component includes the third-order harmonic components g_3,1(n), the third intermodulation beat components g_3,2(n), the triple beat components g_3,3(n), the self-compression/expansion components g_3,4(n), and the cross-compression/expansion components g_3,5(n).

This gives

$k_{3} x^{3} (n) = \sum_{i = 1}^{5} g_{3, i} (n)$

$Where$

$g_{3, 1} (n) = \frac{1}{4} \sum_{i = 1}^{3} A_{i}^{3} \cos (3 a_{i}) / 4$

$g_{3, 2} (n) = \frac{3}{4} \sum_{i = 1}^{3} \sum_{j \neq i} A_{i}^{2} A_{j} \cos (2 a_{i} \pm 2 a_{j})$

$g_{3, 3} (n) = \frac{3}{2} (\prod_{i = 1}^{3} A_{i}) \cos (\sum_{i = 1}^{3} (\pm a_{i}))$

$g_{3, 4} (n) = \frac{3}{4} \sum_{i = 1}^{3} A_{i}^{3} \cos (a_{i})$

$g_{3, 5} (n) = \frac{3}{2} \sum_{i = 1}^{3} \sum_{j \neq i} A_{i} A_{j}^{2} \cos (a_{i})$

The 4th order component includes the DC components g_4,0, the fourth-order harmonic components g_4,1(n), the fourth intermodulation beat components g_4,2(n), the sum/difference beat components g_4,3(n), the second harmonic components g_4,5(n). This gives

$k_{4} x^{4} (n) = \sum_{i = 0}^{5} g_{4, i} (n)$

where

$g_{4, 0} = \frac{3}{8} \sum_{i = 1}^{3} A_{i}^{4} + \frac{3}{4} \sum_{i = 1}^{3} \sum_{j \neq i} A_{i}^{2} A_{j}^{2}$

$g_{4, 1} = \frac{1}{8} \sum_{i = 1}^{3} A_{i}^{4} \cos (4 a_{i})$

$g_{4, 2} = \frac{4}{8} \sum_{i = 1}^{3} \sum_{j \neq i} A_{i}^{3} A_{j} \cos (3 a_{i} \pm a_{j}) + \frac{1 2}{8} \sum_{i = 1}^{3} A_{i}^{2} (\prod_{j \neq i} A_{j}) \cos (2 a_{i} \sum_{j \neq i} (\pm a_{j}))$

$g_{4, 3} = \frac{6}{4} \sum_{i = 1}^{3} \cos (a_{i} \pm a_{\mod (i + 1, 3)}) (A_{i}^{2} A_{\mod (i + 1, 3)}^{2} + A_{i} A_{\mod (i + 1, 3)}^{3} + A_{i}^{3} A_{\mod (i + 1, 3)} + \prod_{j = 1}^{3} A_{i})$

$g_{4, 4} = \frac{3}{2} \sum_{i = 1}^{3} \cos (2 a_{i}) (A_{i}^{2} \sum_{j = 1}^{3} A_{j}^{2})$

The 5th order component includes the fifth-order harmonic components g_5,1(n), the fifth intermodulation beat components g_5,2(n), the self-compression/expansion components g_5,3(n), the cross-compression/expansion components g_5,4(n), the third harmonic components g_5,5(n), the third intermodulation beat components g_5,6(n), and the triple beat components g_5,7(n). This gives

$k_{5} x^{5} (n) = \sum_{i = 1}^{5} g_{5, i} (n)$

Where

$g_{5, 1} (n) = \frac{1}{5} \sum_{i = 1}^{3} A_{i}^{5} \cos (5 a_{i})$

$g_{5, 2} (n) = \frac{5}{8} \sum_{i = 1}^{3} \sum_{j \neq i} A_{i} A_{j}^{4} \cos (a_{i} \pm 4 a_{j}) + A_{i}^{2} A_{j}^{3} \cos (2 a_{i} \pm 3 a_{j}) + A_{i}^{3} A_{j}^{2} \cos (3 a_{i} \pm 2 a_{j}) + A_{i}^{4} A_{j} \cos (4 a_{i} \pm a_{j})$

$g_{5, 3} (n) = \frac{5}{8} \sum_{i = 1}^{3} A_{i}^{5} \cos (a_{i})$

$g_{5, 4} (n) = \frac{15}{4} \sum_{i = 1}^{3} \cos (a_{i}) (\sum_{j \neq i} (A_{i}^{3} A_{j}^{2} + A_{i} A_{j}^{4}) + A_{i} \prod_{j \neq i} A_{j}^{2})$

$g_{5, 5} (n) = \frac{5}{4} \sum_{i = 1}^{3} \cos (3 a_{i}) (A_{i}^{3} \sum_{j = 1}^{3} A_{j}^{2})$

$g_{5, 6} (n) = \frac{15}{4} \sum_{i = 1}^{3} \sum_{j \neq i} \cos (2 a_{i} \pm a_{j}) \times (A_{i}^{3} A_{j}^{2} + A_{i}^{4} A_{j} + A_{i}^{2} A_{j} \prod_{k \neq i, j} A_{k}^{2}))$

$g_{5, 7} (n) = \frac{15}{8} \cos (\sum_{i = 1}^{3} (\pm a_{i})) (\sum_{i = 1}^{3} A_{i} \prod_{j \neq i} A_{j}^{2} + \sum_{i = 1}^{3} A_{i}^{3} \prod_{j \neq i} A_{j})$

These nonlinear spectrum growth expressions can be similarly applied if the signal x(n) is the QAM or OFDM signal. Especially, the harmonics provides us a way to design the input signal vectors for DNN equalizers. Note that some of the spectrums that are deviated too much from the transmitted signal bandwidth will be attenuated by the receiver bandpass filters.

DNN-Based Nonlinear Equalization
A. Nonlinear Equalizer Models

To mitigate the PA nonlinear distortions, nonlinear equalizers can be applied at the receivers. Obviously, the Volterra series model can still be used to analyze the response of nonlinear equalizers. One of the differences from (2) is that the even order nonlinearity may still be included and may increase the nonlinear mitigation effects [5].

Consider the system block diagram of nonlinear equalization shown in FIG. 1, which shows a signal x(n) entering a nonlinear power amplifier, to produce a distorted signal y(n), which passes through a channel h₁, which produces a response r(n), which is fed to a neural network equalizer to produce a corrected output z(n).

Let the received signal be

$\begin{matrix} r (n) = \sum_{ℓ = 0}^{L} h_{ℓ} y (n - ℓ) + v (n), & (9) \end{matrix}$

where h custom character is the finite-impulse response (FIR) channel coefficients and v(n) is additive white Gaussian noise (AWGN). With the received sample sequence r(n), a nonlinear equalizer will generate z(n) as the estimated symbols.

If the PA has only slight nonlinearity as modeled by the simple “AM-AM AM-PM” model (1), the received samples r(n) may be stacked together into M+1 dimensional vectors r(n)=[r(n), . . . , r(n−M)]^T, where (⋅)^Tdenotes transpose, and write the received samples in vector form as

r(n)=HG(n)x(n)+v(n) (10)

where H is an (M+1)×(M+L+1) dimensional channel matrix

$\begin{matrix} H = [\begin{matrix} h_{o} & \dots & h_{L} \\ ⋱ & ⋱ \\ h_{o} & \dots & h_{L} \end{matrix}] & (11) \end{matrix}$

$and$

$G (n) = diag {V_{y (n)} e^{j ψ_{y (n)}}, \dots, V_{n (y - M - L)} e^{j ψ_{y (n - M - L)}}}$

is an (M+L+1)×(M+L+1) dimensional diagonal matrix which consists of the nonlinear PA responses, x(n)=[x(n), . . . , x(n−M−L)]^T, and v(n)=[v(n, . . . , v(n−M)]^T. To equalize the received signal, we apply a nonlinear equalizer with the form

ƒ^T=G′(n)[ƒ₀, . . . ,ƒ_M] (12)

where [ƒ₀, . . . , ƒ_M]H≈[0, . . . , 1, . . . , 0] is to equalize the propagation channel, and

$G^{'} (n) \approx \frac{1}{V_{y (n - d)}} e^{- j ψ_{y (n - d)}}$

is to equalize the nonlinear PA response. Let {circumflex over (r)}(n) be the output of the first linear equalization step. The second nonlinear equalization step can be implemented as a maximum likelihood estimation problem, i.e.,

$z (n) = \arg \min_{\forall x (n)} {❘ \hat{r} (n) - V_{y} e^{j ψ_{y}} x (n) ❘}^{2} .$

This gives the output

custom character (n)=ƒ^Tr(n)≈x(n−d) (13)

with certain equalization delay d.

Both the channel coefficients h custom character and the nonlinear PA responses V_y, ψ_ycan be estimated via training, as can the channel equalizer ƒ^T. Because the PA nonlinearity is significant for large signal amplitude only, we can apply small-amplitude training signals x(n) first to estimate the channel h and the channel equalizer [ƒ₀, . . . , ƒ_M]. We can then remove the channel H from (10) with the first step linear channel equalization. Because the matrix G(n) is diagonal, we can easily estimate G(n) with regular training and then estimate the transmitted symbols as outlined in (13).

For more complex nonlinear PA responses, such as (2), we can conduct channel equalization similarly as (12). First, we can still apply small-amplitude training signals to estimate [ƒ₀, . . . , ƒ_M] so as to equalize the channel h custom character . This linear channel equalization step gives {circumflex over (r)}(n)≈y(n). We can then focus on studying the equalization of nonlinear distortion of the PA, which can in general be conducted with the maximum likelihood method,

$\begin{matrix} {\hat{x} (n) : n = 0, \dots, N} = \arg \min_{{x (n)}} \sum_{n = 0}^{N} {❘ \hat{r} (n) - \hat{y} (n) ❘}^{2}, & (14) \end{matrix}$

where {circumflex over (r)}(n) is the sequence after the linear channel equalization, ŷ(n) is the sequence reconstructed by using the sequence x(n) and the nonlinear PA response parameters b_kdbased on (2), and N is the total number of symbols. The optimization problem (14) can be solved with the Viterbi sequence estimation algorithm if the memory length of the PA is small enough and the PA nonlinear response is known to the receiver.

In case the PA nonlinear response cannot be estimated, the equalization of nonlinear PA response is challenging. In this case, one of the ways is to use the conventional Volterra series equalizer, which approximates G′(n) with a Volterra series model. Similar to (2), this gives

$\begin{matrix} 𝓏 (n) = \sum_{d = 0}^{D} \sum_{k = 0}^{P} g_{kd} \hat{r} (n - d) {❘ \hat{r} (n - d) ❘}^{k - 1} . & (15) \end{matrix}$

The objective of the Volterra series equalizer design is to design g_kdsuch that custom character (n)≈x(n−) for some equalization delay .

Similarly, as the DPD design of [5], based on the Volterra series model (15), we can estimate the coefficients g_kdby casting the estimation into a least squares problem

$\begin{matrix} \min_{{g_{kd}}} \sum_{n = L}^{N} {❘ x (n - L) - \sum_{d = 0}^{D} \sum_{k = 1}^{P} g_{kd} \hat{r} (n - d) {❘ \hat{r} (n - d) ❘}^{k - 1} ❘}^{2}, & (16) \end{matrix}$

with training symbols x(n) and received samples {circumflex over (r)}(n). Note that only the coefficients g_kdare needed to be estimated, and these coefficients are linear with respect to {circumflex over (r)}(n) and x(n).

Define the vector a=[g₀₀, g₀₁, . . . , g_PD]^T, and the vector x=[x(0); . . . , x(N−L)]^T. Define the (N−L+1)×DP data matrix

$\begin{matrix} B = [\begin{matrix} \hat{r} (L) & \hat{r} (L) ❘ \hat{r} (L) ❘ & \dots & {\hat{r} (L - D) ❘ \hat{r} (L - D) ❘}^{P - 1} \\ ⋮ & ⋮ \\ \hat{r} (N) & \hat{r} (N) ❘ \hat{r} (N) ❘ & \dots & \hat{r} (N - D) {❘ \hat{r} (N - D) ❘}^{P - 1} \end{matrix}] & (17) \end{matrix}$

Then, (16) becomes

$\begin{matrix} \min_{a} { x - Ba }^{2} & (18) \end{matrix}$

Solution to (18) is

a=B⁺x (19),

where B⁺=(B^HB)⁻¹B is the pseudo-inverse of the matrix B. From (19), we can obtain the Volterra series equalizer coefficients g_kd. One of the major problems for the Volterra series equalizer is that it is hard to determine the order sizes, i.e., the values of D and P. Even for a nonlinear PA with slight nonlinear effects (i.e., small D and P in (2)), the length of D and P for Volterra series equalizer may be extremely long in order for (15) to have sufficient nonlinearity mitigation capability.

A potential way to resolve this problem is to apply artificial neural networks to fit the nonlinear equalizer response (15). Neural networks can fit arbitrary nonlinearity and can realize this with potentially small sizes. Nevertheless, in conventional neural network equalizers such as [14] [15], the input (features) to the neural networks was simply a time-delayed vector [r(n), . . . , r(n−M)]. Although neural networks may have the capability to learn the nonlinear effects specified in (15), in practice the training may not necessarily converge to the desirable solutions due to local minimum and limited training data. In addition, conventional neural network equalizers were all feed-forward networks with fully connected layers only, which often suffer from problems like shallow network architecture and over-fitting.

It is therefore an object to provide a radio receiver, comprising: an input configured to receive a transmitted radio frequency signal representing a set of symbols communicated through a communication channel; a Volterra series processor configured to decompose the transmitted radio frequency signal as a Volterra series expansion; an equalizer, comprising a deep neural network trained with respect to channel distortion, receiving the Volterra series expansion; and an output, configured to present data corresponding to a reduced distortion of the received distorted transmitted radio frequency signal.

It is also an object to provide a radio reception method, comprising: receiving a transmitted radio frequency signal representing a set of symbols communicated through a communication channel; decomposing the transmitted radio frequency signal as a Volterra series expansion; equalizing the Volterra series expansion with a deep neural network trained with respect to channel distortion, receiving the Volterra series expansion; and presenting data corresponding to a reduced distortion of the received transmitted radio frequency signal.

It is a further object to provide an equalization method for a radio signal, comprising: storing parameters for decomposition of a received radio frequency signal as a Volterra series expansion; processing the Volterra series expansion in a deep neural network comprising a plurality of neural network hidden layers and at least one fully connected neural network layer, trained with respect to radio frequency channel distortion; and presenting an output of the deep neural network. The method may further comprise demodulating the output of the deep neural network, wherein a bit error rate of the demodulator is reduced with respect to an input of the received radio frequency signal to the demodulator.

It is another object to provide an equalizer for a radio receiver, comprising: a memory configured to store parameters for decomposition of a received radio frequency signal as a Volterra series expansion; a deep neural network comprising a plurality of neural network hidden layers and at least one fully connected neural network layer, trained with respect to radio frequency channel distortion, receiving the Volterra series expansion of the received radio frequency signal; and an output configured to present an output of the deep neural network. The system may further comprise a demodulator, configured to demodulate the output, wherein a bit error rate of the demodulator is reduced with respect to an input of the received radio frequency signal to the demodulator.

The Volterra series expansion may comprise at least third or fifth order terms.

The deep neural network may comprise at least two or three convolutional network layers. The deep neural network may comprise at least three one-dimensional convolutional network layers. The convolutional layers may be hidden layers. The deep neural network may comprise at least three one-dimensional layers, each layer having at least 10 feature maps. The radio receiver may further comprise a fully connected layer subsequent to the at least three layers.

The distorted transmitted radio frequency signal comprises an orthogonal frequency multiplexed (OFDM) signal, a quadrature amplitude multiplexed (QAM) signal, a QAM-16 signal, a QAM-64 signal, a QAM-256 signal, a quadrature phase shift keying (QPSK) signal, a 3G signal, a 4G signal, a 5G signal, a WiFi (IEEE-802.11 standard family) signal, a Bluetooth signal, a cable broadcast signal, an optical transmission signal, a satellite radio signal, etc.

The radio receiver may further comprise a demodulator, configured to demodulate output as the set of symbols.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 shows a system block diagram with nonlinear power amplifier and deep neural network equalizer.

FIG. 2 shows a block diagram of DNN equalizer.

FIGS. 3A-3D show constellations of 16 QAM over a simulated PA. FIG. 3A: received signal. FIG. 3B: Volterra equalizer output. FIG. 3C: time-delayed NN output. FIG. 3D: Volterra+NN output.

FIGS. 4A-4D show constellation of 16 QAM over a real PA. FIG. 4A: received signal. FIG. 4B: Volterra equalizer output. FIG. 4C: time-delayed NN output. FIG. 4D: Volterra+NN output.

FIG. 5 shows a comparison of three equalization methods for 16-QAM under various NLD levels.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Volterra-Based DNN Equalizer

The present technology therefore employs deep neural networks to implement the nonlinear equalizer in the receiver, which can mitigate the nonlinear effects of the received signals due to not only PAs but also nonlinear channels and propagations. The architecture of the DNN equalizer is shown in FIG. 2, which shows an input X, which undergoes a series of three 1-d convolutions, and FC (fully-connected) dropout, to produce the output Y.

Different from [10], multi-layer convolutional neural networks (CNNs) are employed. Different from conventional neural network predistorters proposed in [6], neural networks are used as equalizers at the receivers. Different from conventional neural network equalizers such as those proposed in [14] [15], in the present DNN equalizer, not only the linear delayed samples r(n), but also the CNN and the input features in X are used. The Volterra series models are applied to create input features.

We can assume that the linear channel H has already been equalized by a linear equalizer, whose output signal is r(n). In fact, this equalization is not required, but simplifies the presentation of the analysis.

According to Volterra series representation of nonlinear functions, the input-output response of the nonlinear equalizer can be written as

$\begin{matrix} 𝓏 (n) = \sum_{k = 1}^{P} \sum_{d_{1} = 0}^{D} \dots \sum_{d_{k} = 0}^{D} f_{d_{1}, \dots, d_{k}} \prod_{i = 1}^{k} r (n - d_{i}) . & (20) \end{matrix}$

One of major problems is that the number of coefficients ƒ_d₁_{, . . . , d}_kincreases exponentially with the increase of memory length D and nonlinearity order P. There are many different ways to develop more efficient Volterra series representations with reduced number of coefficients. For example, [23], exploits the fact that higher-order terms do not contribute significantly to the memory effects of PAs to reduce the memory depth d when the nonlinearity order k increases.

This technique can drastically reduce the total number of coefficients. In [24] [25] and [26], a dynamic deviation model was developed to reduce the full Volterra series model (20) to the following simplified one:

$𝓏 (n) = 𝓏_{s} (n) + 𝓏_{d} (n) = \sum_{k = 1}^{P} f_{k, 0} r^{k} (n) + \sum_{k = 1}^{P} \sum_{j = 1}^{k} r^{k - j} (n) \sum_{d_{1} = 0}^{D} \dots \sum_{d_{j} = d_{j - 1}}^{D} f_{k, j} \prod_{i = 1}^{j} r (n - d_{i})$

where custom character _s(n) is the static term, and _d(n) is the dynamic term that includes all the memory effects. We can see that the total number of coefficients can be much reduced by controlling the dynamic order j which is a selectable parameter.

We construct the input features of the DNN based on the model (21). Corresponding to the static term custom character _s(n), we change it to:

${\hat{𝓏}}_{s} (n) = \sum_{1 \leq k \leq P} f_{k, 0} r (n) {❘ r (n) ❘}^{k - 1} .$

The reason that (22) changes r^k(n) to r(n)|r(n)|^k-1is that only the signal frequency within the valid passband is interested. This means the input feature vector X should include terms r(n)|r(n)|^k-1. Similarly, corresponding to the dynamic term custom character _d(n), we need to supply r^k-j(n)Π_i−1^jr(n−d_i) in the features where half of the terms r(n) and r(n−d_i) should be conjugated. For simplicity, in the DNN equalizer, the vector X includes r(n−q)|r(n−q)|^k-1for some q and k.

By applying Volterra series components directly as features of the input X, the DNN can develop more complex nonlinear functions with a fewer number of hidden layers and a fewer number of neurons. This will also make the training procedure converge much faster with much less training data.

In FIG. 2, the input X is a tensor formed by the real and imaginary parts of r(n−q)|r(n|q)^k-1with appropriate number of delays q and nonlinearities k. There are three single dimension convolutional layers, each with 20 or 10 feature maps. After a dropout layer for regularization, this is followed by a fully-connected layer with 20 neurons. Finally, there is a fully-connected layer to form the output tensor Y which has two dimensions. The output Y is used to construct the complex custom character (n), where (n)={circumflex over (x)}(n−d) for some appropriate delay d. All the convolutional layers and the first fully connected layer use the sigmoid activation function, while the output layer uses the linear activation function. The mean square error loss function L^loss=E[|x(n−d)− custom character (n)|²] is used, where (n) is replaced by Y and x(n−d) is replaced by training data labels.

Experiment Evaluations

Experiments are presented on applying the Volterra series based DNN equalizer (Volterra+NN) for nonlinear PA equalization. The (Volterra+NN) scheme with the following equalization methods: a Volterra series-based equalizer (Volterra) and a conventional time-delay neural network equalizer (NN). The performance metrics are mean square error (MSE)

√{square root over (E[|z(n)−x(n−d)|²]/E[|x(n−d)|²])}

and symbol error rate (SER).

Both simulated signals and real measurement signals were employed. To generate simulated signals, a Doherty nonlinear PA model consisting of 3rd and 5th order nonlinearities was employed. Referring to (2), the coefficients b_k,qwere

b_0,0:2={1.0513+0.0904j,−0.068−0.0023j,0.0289−0.0054j}
b_2,0:2={−0.0542−0.29j,0.2234+0.2317j,−0.0621−0.0932j}
b_4,0:2={−0.9657−0.7028j,−0.2451−0.3735j,0.1229+0.1508j},

which was used in [5] to simulate a 5th order dominant nonlinear distortion derived from PA devices used in the satellite industry. For real measurement, our measurement signals were obtained from PA devices used in the cable TV (CATV) industry, which are typically dominated by 3^rdorder nonlinear distortion (NLD). Various levels of nonlinear distortion, in terms of dBc, were generated by adjusting the PAs.

For the Volterra equalizer, the approximate response of the nonlinear equalizer with delays including 8 pre- and post-main taps and with nonlinearities including even and odd order nonlinearity up to the 5th order was employed. To determine the values of the Volterra coefficients, N=4; 096 training symbols were transmitted through the PA and then collected the noisy received samples r(n).

For the conventional time-delay NN equalizer, a feedforward neural network with an 80-dimensional input vector X and 5 fully-connected hidden layers with 20, 20, 10, 10, 10 neurons, respectively, was applied.

FIG. 3 shows the constellation and MSE of the equalizer's outputs. It can be seen that the proposed scheme provides the best performance.

FIG. 4 shows the constellation of 16 QAM equalization over the real PA. The corresponding SER were 0.0067, 0.0027, 0.00025, respectively. It can be seen that the Volterra+NN scheme has the best performance.

FIG. 5 provides MSE measurements for 16-QAM under various nonlinear distortion level dBc. For each 1 dB increase in NLD, the resultant MSE is shown for the “Measured”, “Volterra”, “NN”, and the proposed “Volterra+NN” cases. MSE reduction diminishes appreciably as modulation order increases from QPSK to 64-QAM, but small improvements in MSE have been observed lead to appreciable SER improvement, especially for more complex modulation orders. The 4,096 symbol sample sizes have limited the measurements to a minimum measurable 0.000244 SER, which represents 1 symbol error out of 4,096 symbols.

Table 2 summarizes equalization performance, which shows the averaged percent reduction/improvement in MSE and SER from the NLD impaired data for multiple modulation orders. Note that 0% SER improvement for QPSK was because the received signal's SER was already very low.

TABLE 2

EQ-MSE/SER improvement in percentage over measured NLD

Volterra
NN
Volterra + NN

MSE
SER
MSE
SER
MSE
SER

64-QAM
16%
26%
10%
25%
42%
44%

16-QAM
41%
2%
35%
6%
85%
28%

QPSK
57%
0%
100%
0%
100%
0%

Average
38%
9%
48%
10%
76%
24%

The nonlinear equalization scheme presented by integrating the Volterra series nonlinear model with deep neural networks yields superior results over conventional nonlinear equalization approaches in mitigating nonlinear power amplifier distortions. It finds application for many 5G communication scenarios.

The technology may be implemented as an additional component in a receiver, or within the digital processing signal chain of a modern radio. A radio is described in US 20180262217, expressly incorporated herein by reference.

In an implementation, a base station may include a SDR receiver configured to allow the base station to operate as an auxiliary receiver. In an example implementation, the base station may include a wideband receiver bank and a digital physical/media access control (PHY/MAC) layer receiver. In this example, the SDR receiver may use a protocol analyzer to determine the protocol used by the source device on the uplink to the primary base station, and then configure the digital PHY/MAC layer receiver for that protocol when operating as art auxiliary receiver. Also, the digital PHY/MAC layer receiver may be configured to operate according to another protocol when operating as a primary base station. In another example, the base station may include a receiver hank for a wireless system, for example, a fifth Generation (5G) receiver bank, and include an additional receiver having SDR configurable capability. The additional receiver may be, for example, a digital Wi-Fi receiver configurable to operate according to various Wi-Fi protocols. The base station may use a protocol analyzer to determine the particular Wi-Fi protocol used by the source device on the uplink to the primary base station. The base station may then configure the additional receiver as the auxiliary receiver for that Wi-Fi protocol.

Depending on the hardware configuration, a receiver may be used to flexibly provide uplink support in systems operating according to one or more protocols such as the various IEEE 802.11 Wi-Fi protocols, 3^rdGeneration Cellular (3G), 4^thGeneration Cellular (4G) wide band code division multiple access (WCDMA), Long Term Evolution (LTE) Cellular, and 5^thgeneration cellular (5G).

See, 5G References, infra.

Processing unit may comprise one or more processors, or other control circuitry or any combination of processors and control circuitry that provide, overall control according to the disclosed embodiments. Memory may be implemented as any type of as any type of computer readable storage media, including non-volatile and volatile memory.

The example embodiments disclosed herein may be described in the general context of processor-executable code or instructions stored on memory that may comprise one or more computer readable storage media (e.g., tangible non-transitory computer-readable storage media such as memory). As should be readily understood, the terms “computer-readable storage media” or “non-transitory computer-readable media” include the media for storing of data, code and program instructions, such as memory, and do not include portions of the media for storing transitory propagated or modulated data communication signals.

While the functionality disclosed herein has been described by illustrative example using descriptions of the various components and devices of embodiments by referring to functional blocks and processors or processing units, controllers, and memory including instructions and code, the functions and processes of the embodiments may be implemented and performed using any type of processor, circuit, circuitry or combinations of processors and or circuitry and code. This may include, at least in part, one or more hardware logic components. For example, and without limitation, illustrative types of hardware logic components that can be used include field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-a-chip systems (SOCs), complex programmable logic devices (CPLDs), etc. Use of the term processor or processing unit in this disclosure is mean to include all such implementations.

The disclosed implementations include a receiver, one or more processors in communication with the receiver, and memory in communication with the one or more processors, the memory comprising code that, when executed, causes the one or more processors to control the receiver to implement various features and methods according to the present technology.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example embodiments, implementations, and forms of implementing the claims and these example configurations and arrangements may be changed significantly without departing from the scope of the present disclosure. Moreover, although the example embodiments have been illustrated with reference to particular elements and operations that facilitate the processes, these elements, and operations may be combined with or, be replaced by, any suitable devices, components, architecture or process that achieves the intended functionality of the embodiment. Numerous other changes, substitutions, variations, alterations, and modifications may be ascertained to one skilled in the art and it is intended that the present disclosure encompass all such changes, substitutions, variations, alterations, and modifications a falling within the scope of the appended claims.

REFERENCES

[1] J.-A. Lucciardi, P. Potier, G. Buscarlet, F. Barrami, and G. Mesnager, “Non-linearized amplifier and advanced mitigation techniques: Dvbs-2× spectral efficiency improvement,” in GLOBECOM 2017-2017 IEEE Global Communications Conference. IEEE, 2017, pp. 1-7.

[2] J. Wood, Behavioral modeling and linearization of RF power amplifiers. Artech House, 2014.

[3] C.-L. Wang and Y. Ouyang, “Low-complexity selected mapping schemes for peak-to-average power ratio reduction in ofdm systems,” IEEE Transactions on signal processing, vol. 53, no. 12, pp. 4652-4660, 2005.

[4] J. Kim and K. Konstantinou, “Digital predistortion of wideband signals based on power amplifier model with memory,” Electronics Letters, vol. 37, no. 23, pp. 1417-1418, 2001.

[5] L. Ding, G. T. Zhou, D. R. Morgan, Z. Ma, J. S. Kenney, J. Kim, and C. R. Giardina, “A robust digital baseband predistorter constructed using memory polynomials,” IEEE Transactions on communications, vol. 52, no. 1, pp. 159-165, 2004.

[6] M. Rawat, K. Rawat, and F. M. Ghannouchi, “Adaptive digital predistortion of wireless power amplifiers/transmitters using dynamic realvalued focused time-delay line neural networks,” IEEE Transactions on Microwave Theory and Techniques, vol. 58, no. 1, pp. 95-104, 2010.

[7] S. Dimitrov, “Non-linear distortion cancellation and symbol-based equalization in satellite forward links,” IEEE Trans Wireless Commun, vol. 16, no. 7, pp. 4489-4502, 2017.

[8] D. J. Sebald and J. A. Bucklew, “Support vector machine techniques for nonlinear equalization,” IEEE Transactions on Signal Processing, vol. 48, no. 11, pp. 3217-3226, 2000.

[9] S. Chen, B. Mulgrew, and P. M. Grant, “A clustering technique for digital communications channel equalization using radial basis function networks,” IEEE Transactions on neural networks, vol. 4, no. 4, pp. 570-590, 1993.

[10] B. Li, C. Zhao, M. Sun, H. Zhang, Z. Zhou, and A. Nallanathan, “A bayesian approach for nonlinear equalization and signal detection in millimeter-wave communications,” IEEE Transactions on Wireless Communications, vol. 14, no. 7, pp. 3794-3809, 2015.

[11] F. Mkadem and S. Boumaiza, “Physically inspired neural network model for rf power amplifier behavioral modeling and digital predistortion,” IEEE Transactions on Microwave Theory and Techniques, vol. 59, no. 4, pp. 913-923, 2011.

[12] T. Liu, S. Boumaiza, and F. M. Ghannouchi, “Dynamic behavioral modeling of 3 g power amplifiers using real-valued time-delay neural networks,” IEEE Transactions on Microwave Theory and Techniques, vol. 52, no. 3, pp. 1025-1033, 2004.

[13] M. Ibnkahla, “Applications of neural networks to digital communications-a survey,” Signal processing, vol. 80, no. 7, pp. 1185-1215, 2000.

[14] D.-C. Park and T.-K. J. Jeong, “Complex-bilinear recurrent neural network for equalization of a digital satellite channel,” IEEE Transactions on Neural Networks, vol. 13, no. 3, pp. 711-725, 2002.

[15] A. Uncini, L. Vecci, P. Campolucci, and F. Piazza, “Complex-valued neural networks with adaptive spline activation function for digital-radio-links nonlinear equalization,” IEEE Transactions on Signal Processing, vol. 47, no. 2, pp. 505-514, 1999.

[16] M. S. Sim, M. Chung, D. Kim, J. Chung, D. K. Kim, and C.-B. Chae, “Nonlinear self-interference cancellation for full-duplex radios: From link-level and system-level performance perspectives,” IEEE Communications Magazine, vol. 55, no. 9, pp. 158-167, 2017.

[17] I. Yoffe and D. Wulich, “Predistorter for mimo system with nonlinear power amplifiers,” IEEE Transactions on Communications, vol. 65, no. 8, pp. 3288-3301, 2017.

[18] M. Abdelaziz, L. Anttila, and M. Valkama, “Reduced-complexity digital predistortion for massive mimo,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 6478-6482.

[19] H. Yan and D. Cabric, “Digital predistortion for hybrid precoding architecture in millimeter-wave massive mimo systems,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 3479-3483.

[20] C. Mollén, E. G. Larsson, and T. Eriksson, “Waveforms for the massive mimo downlink: Amplifier efficiency, distortion, and performance,” IEEE Transactions on Communications, vol. 64, no. 12, pp. 5050-5063, 2016.

[21] A. Cheaito, M. Crussière, J.-F. Helard, and Y. Louët, “Quantifying the memory effects of power amplifiers: Evm closed-form derivations of multicarrier signals.” IEEE Wireless Commun. Letters, vol. 6, no. 1, pp. 34-37, 2017.

[22] K. Simons, Technical Handbook for CATV Systems, 3rd Edition. Jerrod Publication No. 436-001-01, 1968.

[23] J. Staudinger, J.-C. Nanan, and J. Wood, “Memory fading volterra series model for high power infrastructure amplifiers,” in Radio and Wireless Symposium (RWS), 2010 IEEE. IEEE, 2010, pp. 184-187.

[24] A. Zhu, J. C. Pedro, and T. J. Brazil, “Dynamic deviation reduction-based volterra behavioral modeling of rf power amplifiers,” IEEE Transactions on microwave theory and techniques, vol. 54, no. 12, pp. 4323-4332, 2006.

[25] A. Zhu, P. J. Draxler, J. J. Yan, T. J. Brazil, D. F. Kimball, and P. M. Asbeck, “Open-loop digital predistorter for rf power amplifiers using dynamic deviation reduction-based volterra series,” IEEE Transactions on Microwave Theory and Techniques, vol. 56, no. 7, pp. 1524-1534, 2008.

[26] L. Guan and A. Zhu, “Simplified dynamic deviation reduction-based volterra model for doherty power amplifiers,” in Integrated Nonlinear Microwave and Millimetre-Wave Circuits (INMMIC), 2011 Workshop on. IEEE, 2011, pp. 1-4.

[27] Schetzen, M. The Volterra and Wiener Theories of Non-linear Systems. (1980) Wiley & Sons.

[28] Black, H. S. [October, 1928] Translating system. U.S. Pat. No. 1,686,792.

[29] Black, H. S. [December, 1937] Wave translating system. U.S. Pat. No. 2,102,671.

[30] Mitchell, A. F. [November, 1979] A 135 MHz feedback amplifier. IEEE Colloq. Broadband High Frequency Amplifiers.

[31] Arthanayake, T. and Wood, H. B. [8 Apr. 1971] Linear amplification using envelope feedback. Elec. Lett.

[32] Chadwick, P. [1986] Wideband Amplifier Applications Book, Edition 2, Plessey Semiconductor.

Amplifier References

Aghvami, A. H. and Robertson, I. D. [April, 1993] Power limitation and high-power amplifier non linearities in on-board satellite communications systems. Electron. and Comm. Engin. J.

Arthanayake, T. and Wood, H. B. [8 Apr. 1971] Linear amplification using envelope feedback. Elec. Lett.

Bennet, T. J. and Clements, R. F. [May, 1974] Feedforward—An alternative approach to amplifier linearisation. Radio and Electron. Engin.

Bhargava, V. K. et al. [1981] Digital Communications by Satellite, John Wiley and Sons.

Black, H. S. [December, 1937] Wave translating system. U.S. Pat. No. 2,102,671.

Black, H. S. [October, 1928] Translating system. U.S. Pat. No. 1,686,792.

Bond F. E. and Meyer, H. F. [April, 1970] Intermodulation effects in limiter amplifier repeaters. IEEE Trans. Comm., Vol. COM-18, p. 127-135.

Chadwick, P. [1986] Wideband Amplifier Applications Book, Edition 2, Plessey Semiconductor.

Cole, R. A. [December, 1989] Linearisation of a power amplifier using Cartesian Loop feedback. Report No. 72/89/R/451/C. Roke Manor Res.

ETSI [August, 1994] Standard ETR 132. Radio broadcasting systems; Code of practice for site engineering VHF FM sound broadcasting transmitters. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.

ETSI [January, 1995] European Standard ETS 300 384. Radio broadcasting systems; Very high frequency (VHF), frequency modulated, sound broadcasting transmitters. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.

ETSI [June, 1998] Standard ETR 053 Ed 3—Radio site engineering for equipment and systems in the mobile service. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.

ETSI [March, 1997] European Standard ETS 300 113. Radio equipment and systems (RES); Land mobile service; Technical characteristics and test conditions for radio equipment intended for the transmission of data (and speech) and having an antenna connector. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.

Gray, L. F. [1980] Application of broadband linearisers to satellite transponders. IEEE Conf. Proc. ICC'80.

Heathman, A. C. [1989] Methods for intermodulation prediction in communication systems. Ph. D. Thesis, University of Bradford, United Kingdom.

IESS [November, 1996] IESS-401 (Rev. 4). Performance requirements for intermodulation products transmitted from INTELSAT earth stations. Intelsat Earth Station Standard (IESS).

Kaeadar, K. [December, 1986] Gaussian white-noise generation for digital signal synthesis. IEEE Trans. Inst. and Meas., Vol. IM 35, 4.

Kahn, L. R. [July, 1952] SSB transmission by envelope elimination and restoration. Proc. IRE.

Mitchell, A. F. [November, 1979] A 135 MHz feedback amplifier. IEEE Colloq. Broadband High Frequency Amplifiers.

Pavliouk, A. [1977] Unification of measurement procedures for out-of-band emission spectra and peak envelope power of single-sideband radio transmitter measurements. Proc. of the NIIR, 4 (in Russian).

Petrovic, V. and Gosling, W. [10 May 1979] Polar loop transmitter. Elec. Lett.

Pye Telecom [November, 1978] Intermodulation in VHF and UHF radio systems—locating and minimizing the effects. Engineering Notes, Pub. Ref. No. TSP480/1, United Kingdom.

Radiocommunications Agency [April, 1987] Code of practice for radio site engineering. MPT 1331. Radiocommunications Agency (RA), Flyde Microsystems Ltd. United Kingdom.

Saleh, A. M. [May 1982] Intermodulation analysis of FDMA satellite systems employing compensated and uncompensated TWT‘s’. IEEE Trans. Comm., Vol. COM-30, 5.

Schetzen, M. [1980] The Volterra and Wiener Theories of Non-linear Systems. Wiley & Sons.

Schetzen, M. [1980] The Volterra and Wiener Theories of Non-linear Systems. Wiley & Sons.

Shahid, M., Shepherd, S. J., Lin, B., Khairruddin, I., and Barton, S. K. [December, 1996] Study of methods of measuring multi-carrier intermodulation performance Report No. 581, Purchase Order No. 142379 d′ESA, with University of Bradford, United Kingdom.

Shimbo, O. [February, 1971] Effects of intermodulation, AM-PM conversion, and additive noise in multicarrier TWT systems. Proc. IEEE, Vol. 59, p. 230-238.

Smith, C. N. [1986] Application of the polar loop technique to UHF SSB transmitters. Ph.D. Thesis, University of Bath.

Smith, C. N. and PETROVIC, V. [1982] Cartesian loop transmitter. Internal Research Report, University of Bath, School of Electrical and Electronic Engineering.

Tondryk, W. [1991] Intermodulation testing of the INMARSAT payload—Response to system PDR. Marconi Space Systems Ltd.

Wassermann, M. et al. [1983] Study and breadboarding of an L-band high power linearized TWT amplifier. Final Report, ESTEC contract No. 5459/83/NL/GM.

Wood, A. [October 1998] Radio interference: Sources and solutions. LPRA NEWS, p. 21.

Volterra Series References

Bohm, D. The Special Theory of Relativity, Benjamin, 1965.

Censor, D., & Melamed, T, 2002, Volterra differential constitutive operators and locality considerations in electromagnetic theory, PIER—Progress in Electromagnetic Research, 36: 121-137

Censor, D., 2000, A quest for systematic constitutive formulations for general field and wave systems based on the Volterra differential operators, PIER—Progress In Electromagnetics Research, (25): 261-284

Censor, D., 2001, Constitutive relations in inhomogeneous systems and the particle-field conundrum, PIER—Progress In Electromagnetics Research, (30): 305-335

Schetzen, M., 1980, The Volterra and Wiener Theorems of Nonlinear Systems, New York, Chichester, Brisbane and Toronto: John Wiley and Sons

Sonnenschein, M & Censor, D., 1998, Simulation of Hamiltonian light beam propagation in nonlinear media, JOSA—Journal of the Optical Society of America B, (15): 1335-1345

Filter References

Akaiwa, Y. Introduction to Digital Mobile Communication. New York: Wiley, 1997.

Altera Corporation. Digital Predistortion Reference Design. Application Note 314, 2003.

Aysal, Tuncer C., and Kenneth E. Barner, “Myriad-Type Polynomial Filtering”, IEEE Transactions on Signal Processing, vol. 55, no. 2, February 2007.

Barner, Kenneth E., and Tuncer Can Aysal, “Polynomial Weighted Median Filtering”, IEEE Transactions on Signal Processing, vol. 54, no. 2, February 2006.

Biglieri, Ezio, Sergio Barberis, and Maurizio Catena, “Analysis and Compensation of Nonlinearities in Digital Transmission Systems”, IEEE Journal on selected areas in Communications, vol. 6, no. 1, January 1988.

Budura, Georgeta, and Corina Botoca, “Efficient Implementation of the Third Order RLS Adaptive Volterra Filter”, FACTA Universitatis (NIS) Ser.: Elec. Energ. vol. 19, no. 1, April 2006.

Ding, L., et al. “A Robust Digital Baseband Predistorter Constructed Using Memory Polynomials,” IEEE Transactions on Communications, Vol. 52, No. 1, June 2004.

Fang, Yang-Wang, Li-Cheng Jiao, Xian-Da Zhang and Jin Pan, “On the Convergence of Volterra Filter Equalizers Using a Pth-Order Inverse Approach”, IEEE Transactions on Signal Processing, vol. 49, no. 8, August 2001.

Guérin, Alexandre, Gérard Faucon, and Régine Le Bouquin-Jeannès, “Nonlinear Acoustic Echo Cancellation Based on Volterra Filters”, IEEE Transactions on Speech and Audio Processing, vol. 11, no. 6, November 2003.

Haykin, Simon, “Adaptive Filter Theory”, Fourth Edition, Pearson Education, 2008.

Kamiya, N., and F. Maehara. “Nonlinear Distortion Avoidance Employing Symbol-wise Transmit Power Control for OFDM Transmission,” Proc. of Int'l. OFDM Workshop, Hamburg, 2009.

Kim, J., and K. Konstantinou. “Digital predistortion of wideband signals based on power amplifier model with memory,” Electronic Letters, Vol. 37, No. 23, November 2001.

Krall, Christoph, Klaus Witrisal, Geert Leus and Heinz Koeppl, “Minimum Mean-Square Error Equalization for Second-Order Volterra Systems”, IEEE Transactions on Signal Processing, vol. 56, no. 10, October 2008.

Leis, John, “Adaptive Filter Lecture Notes & Examples”, Nov. 1, 2008 www.usq.edu.au/users/leis/notes/sigproc/adfilt.pdf.

López-Valcarce, Roberto, and Soura Dasgupta, “Second-Order Statistical Properties of Nonlinearly Distorted Phase-Shift Keyed (PSK) Signals”, IEEE Communications Letters, vol. 7, no. 7, July 2003.

Lozhkin, Alexander N. “Turbo Linearizer for High Power Amplifier.” In 2011 IEEE 73rd Vehicular Technology Conference (VTC Spring), pp. 1-5. IEEE, 2011.

Mathews, V. John, “Adaptive Polynomial Filters,” IEEE Signal Processing Magazine, Vol. 8, No. 3, July 1991.

Park, Dong-Chul, and Tae-Kyun Jung Jeong, “Complex-Bilinear Recurrent Neural Network for Equalization of a Digital Satellite Channel”, IEEE Transactions on Neural Networks, vol. 13, no. 3, May 2002.

Rai, Amrita, and Amit Kumar Kohli. “Analysis of Adaptive Volterra Filters with LMS and RLS Algorithms.” AKGEC Journal of Technology 2, no. 1 (2011).

Therrien, Charles W., W. Kenneth Jenkins, and Xiaohui Li, “Optimizing the Performance of Polynomial Adaptive Filters: Making Quadratic Filters Converge Like Linear Filters”, IEEE Transactions on Signal Processing, vol. 47, no. 4, April 1999.

Tsimbinos John, and Langford B. White, “Error Propagation and Recovery in Decision-Feedback Equalizers for Nonlinear Channels”, IEEE Transactions on Communications, vol. 49, no. 2, February 2001.

Woo, Young Yun, et al. “Adaptive Digital Feedback Predistortion Technique for Linearizing Power Amplifiers,” IEEE Transactions on Microwave Theory and Techniques, Vol. 55, No. 5, May 2007.

Zaknich, A., “Principal of Adaptive Filter and Self Learning System”, Springer Link 2005.

Volterra Series Patents

U.S. Pat. Nos. and Published Patent Application Nos.: U.S. Pat. Nos. 4,615,038; 4,669,116; 4,870,371; 5,038,187; 5,309,481; 5,329,586; 5,424,680; 5,438,625; 5,539,774; 5,647,023; 5,692,011; 5,694,476; 5,744,969; 5,745,597; 5,790,692; 5,792,062; 5,815,585; 5,889,823; 5,924,086; 5,938,594; 5,991,023; 6,002,479; 6,005,952; 6,064,265; 6,166,599; 6,181,754; 6,201,455; 6,201,839; 6,236,837; 6,240,278; 6,288,610; 6,335,767; 6,351,740; 6,381,212; 6,393,259; 6,406,438; 6,408,079; 6,438,180; 6,453,308; 6,504,885; 6,510,257; 6,512,417; 6,532,272; 6,563,870; 6,600,794; 6,633,208; 6,636,115; 6,668,256; 6,687,235; 6,690,693; 6,697,768; 6,711,094; 6,714,481; 6,718,087; 6,775,646; 6,788,719; 6,812,792; 6,826,331; 6,839,657; 6,850,871; 6,868,380; 6,885,954; 6,895,262; 6,922,552; 6,934,655; 6,940,790; 6,947,857; 6,951,540; 6,954,476; 6,956,433; 6,982,939; 6,992,519; 6,999,201; 6,999,510; 7,007,253; 7,016,823; 7,061,943; 7,065,511; 7,071,797; 7,084,974; 7,092,043; 7,113,037; 7,123,663; 7,151,405; 7,176,757; 7,209,566; 7,212,933; 7,236,156; 7,236,212; 7,239,301; 7,239,668; 7,251,297; 7,268,620; 7,272,594; 7,286,009; 7,295,961; 7,304,591; 7,305,639; 7,308,032; 7,333,559; 7,348,844; 7,400,807; 7,403,884; 7,412,469; 7,423,699; 7,436,883; 7,443,326; 7,489,298; 7,512,900; 7,542,518; 7,551,668; 7,570,856; 7,571,401; 7,576,606; 7,589,725; 7,590,518; 7,602,240; 7,606,539; 7,610,183; 7,657,405; 7,720,232; 7,720,236; 7,728,658; 7,729,446; 7,733,177; 7,746,955; 7,755,425; 7,760,887; 7,773,692; 7,774,176; 7,795,858; 7,796,960; 7,808,315; 7,812,666; 7,821,337; 7,821,581; 7,822,146; 7,826,624; 7,847,631; 7,852,913; 7,853,443; 7,864,881; 7,873,172; 7,885,025; 7,885,797; 7,889,007; 7,894,788; 7,895,006; 7,899,416; 7,902,925; 7,903,137; 7,924,942; 7,929,375; 7,932,782; 7,970,150; 7,970,151; 7,979,837; 7,991,073; 7,991,167; 7,995,674; 8,005,858; 8,023,668; 8,031,882; 8,039,871; 8,045,066; 8,046,199; 8,065,060; 8,089,689; 8,105,270; 8,139,630; 8,148,983; 8,149,950; 8,160,191; 8,165,854; 8,170,508; 8,185,853; 8,193,566; 8,195,103; 8,199,399; 8,213,880; 8,244,787; 8,260,732; 8,265,583; 8,270,530; 8,294,605; 8,295,790; 8,306,488; 8,310,312; 8,315,970; 8,331,511; 8,331,879; 8,345,348; 8,346,692; 8,346,693; 8,346,711; 8,346,712; 8,351,876; 8,354,884; 8,355,684; 8,358,169; 8,364,095; 8,369,447; 8,369,595; 8,380,773; 8,390,375; 8,390,376; 8,396,693; 8,410,843; 8,410,850; 8,412,133; 8,421,534; 8,432,220; 8,437,513; 8,463,582; 8,467,438; 8,477,581; 8,483,343; 8,483,450; 8,487,706; 8,489,047; 8,494,463; 8,498,369; 8,509,347; 8,509,712; 8,519,440; 8,532,215; 8,532,964; 8,538,039; 8,564,368; 8,565,343; 8,577,311; 8,587,375; 8,599,050; 8,605,814; 8,605,819; 8,611,190; 8,611,459; 8,611,820; 8,615,208; 8,619,905; 8,620,631; 8,626,089; 8,649,743; 8,675,925; 8,704,595; 8,705,166; 8,712,345; 8,718,178; 8,718,209; 8,724,857; 8,737,937; 8,737,938; 8,744,141; 8,744,377; 8,758,271; 8,761,409; 8,766,917; 8,767,869; 8,780,693; 8,787,628; 8,798,559; 8,804,807; 8,804,871; 8,811,532; 8,823,452; 8,831,074; 8,831,133; 8,831,135; 8,838,218; 8,843,088; 8,843,089; 8,849,611; 8,855,175; 8,855,234; 8,867,601; 8,874,411; 8,885,765; 8,886,341; 8,891,701; 8,896,471; 8,897,351; 8,903,192; 8,909,176; 8,909,328; 8,933,752; 8,934,573; 8,958,470; 8,964,901; 8,964,996; 8,971,834; 8,976,896; 8,994,657; 8,995,571; 8,995,835; 9,008,153; 9,014,299; 9,019,643; 9,020,454; 9,025,607; 9,031,168; 9,036,734; 9,048,865; 9,048,900; 9,071,313; 9,077,508; 9,088,472; 9,094,036; 9,094,151; 9,104,921; 9,106,304; 9,130,628; 9,137,492; 9,143,274; 9,160,280; 9,160,310; 9,160,687; 9,166,610; 9,166,635; 9,166,698; 9,171,534; 9,184,784; 9,185,529; 9,189,458; 9,191,041; 9,191,049; 9,199,860; 9,209,753; 9,209,841; 9,214,968; 9,214,969; 9,225,295; 9,225,501; 9,231,530; 9,231,647; 9,231,801; 9,236,996; 9,246,525; 9,246,731; 9,252,798; 9,252,821; 9,253,608; 9,257,943; 9,258,156; 9,261,978; 9,264,153; 9,265,461; 9,270,304; 9,270,512; 9,271,123; 9,276,602; 9,294,113; 9,304,501; 9,306,606; 9,311,535; 9,312,892; 9,314,623; 9,322,906; 9,337,781; 9,337,783; 9,352,155; 9,361,681; 9,361,936; 9,362,869; 9,362,942; 9,363,068; 9,369,093; 9,369,255; 9,369,541; 9,397,516; 9,404,950; 9,413,516; 9,419,722; 9,431,972; 9,438,178; 9,438,356; 9,439,597; 9,451,920; 9,460,246; 9,461,597; 9,461,676; 9,473,077; 9,479,322; 9,509,331; 9,509,350; 9,517,030; 9,531,475; 9,536,539; 9,537,759; 9,544,126; 9,559,831; 9,564,876; 9,571,312; 9,575,570; 9,590,664; 9,590,668; 9,595,920; 9,595,982; 9,607,003; 9,607,628; 9,608,676; 9,608,718; 9,614,554; 9,628,119; 9,628,120; 9,646,116; 9,647,717; 9,654,211; 9,654,216; 9,659,120; 9,660,593; 9,660,730; 9,665,510; 9,667,292; 9,674,368; 9,680,423; 9,680,497; 9,697,845; 9,705,477; 9,706,296; 9,712,179; 9,712,233; 9,713,010; 9,722,646; 9,722,691; 9,726,701; 9,727,677; 9,735,741; 9,735,800; 9,735,811; 9,735,876; 9,737,258; 9,742,599; 9,746,506; 9,749,161; 9,755,691; 9,762268; 9,768,891; 9,778,902; 9,780,869; 9,780,881; 9,787,459; 9,794,000; 9,800,437; 9,800,734; 9,820311; 9,831,899; 9,837,970; 9,843,346; 9,859,845; 9,866,183; 9,877,265; 9,882,648; 9,887,862; 9,900088; 9,912,435; 9,913,194; 9,923,524; 9,923,640; 9,923,714; 9,928,212; 9,935,590; 9,935,645; 9,935715; 9,935,761; 9,940,938; 9,941,963; 9,953,656; 9,954,384; 9,960,794; 9,960,804; 9,960,900; 9,971920; 9,973,279; 9,974,957; 9,983,243; 9,998,223; 9,998,406; 9,999,780; 10,008,218; 10,009,050; 10,009,109; 10,009,259; 10,013,515; 10,015,593; 10,033,413; 10,033,568; 10,050,636; 10,050,710; 10,050,714; 10,063,265; 10,063,364; 10,075,201; 10,095,927; 10,097,273; 10,097,939; 10,101,370; 10,108,858; 10,110,315; 10,116,390; 10,128,955; 10,141,944; 10,142,754; 10,147,431; 10,148,417; 10,153,793; 10,181,825; 10,224,970; 20010036334; 20010051871; 20020041210; 20020060827; 20020075918; 20020126604; 20020146993; 20020161539; 20020161542; 20020169585; 20020178133; 20020181521; 20020186874; 20030046045; 20030057963; 20030063854; 20030071684; 20030142832; 20030195706; 20030223507; 20040019443; 20040044489; 20040130394; 20040136423; 20040155707; 20040179629; 20040208242; 20040258176; 20050021266; 20050021319; 20050031117; 20050031131; 20050031132; 20050031133; 20050031134; 20050031137; 20050031138; 20050031139; 20050031140; 20050049838; 20050100065; 20050141637; 20050141659; 20050174167; 20050177805; 20050180526; 20050226316; 20050237111; 20050243061; 20050253806; 20050270094; 20050271216; 20050273188; 20060039498; 20060052988; 20060083389; 20060093128; 20060095236; 20060104395; 20060104451; 20060133536; 20060209982; 20060222128; 20060239443; 20060256974; 20060262942; 20060262943; 20060264187; 20060269074; 20060269080; 20060274904; 20070005326; 20070018722; 20070030076; 20070033000; 20070063770; 20070080841; 20070133713; 20070133719; 20070136018; 20070136045; 20070152750; 20070160221; 20070168100; 20070190952; 20070229154; 20070237260; 20070247425; 20070252651; 20070252813; 20070276610; 20080001947; 20080032642; 20080129379; 20080130787; 20080130788; 20080130789; 20080152037; 20080158154; 20080158155; 20080180178; 20080240325; 20080261541; 20080283882; 20080285640; 20080293372; 20090003134; 20090027117; 20090027118; 20090058521; 20090067643; 20090072901; 20090075610; 20090094304; 20090146740; 20090153132; 20090185613; 20090256632; 20090287624; 20090289706; 20090291650; 20090302938; 20090302940; 20090318983; 20100007489; 20100033180; 20100060355; 20100090762; 20100093290; 20100094603; 20100097714; 20100114813; 20100135449; 20100148865; 20100152547; 20100156530; 20100183106; 20100194474; 20100199237; 20100254450; 20100283540; 20100292602; 20100292752; 20100311361; 20100312495; 20110003570; 20110025414; 20110028859; 20110037518; 20110054354; 20110054355; 20110064171; 20110069749; 20110081152; 20110085678; 20110087341; 20110096865; 20110102080; 20110103455; 20110110473; 20110121897; 20110125684; 20110125685; 20110125686; 20110125687; 20110140779; 20110144961; 20110149714; 20110177956; 20110181360; 20110204975; 20110211842; 20110268226; 20110270590; 20110293051; 20120007153; 20120007672; 20120027070; 20120029663; 20120086507; 20120093376; 20120098481; 20120098596; 20120119810; 20120140860; 20120147993; 20120154040; 20120154041; 20120158384; 20120165633; 20120176190; 20120176609; 20120217557; 20120229206; 20120256687; 20120259600; 20120263256; 20120306573; 20120328128; 20130005283; 20130009702; 20130015917; 20130030239; 20130034188; 20130040587; 20130044791; 20130044836; 20130093676; 20130113559; 20130114762; 20130166259; 20130170842; 20130176153; 20130207723; 20130222059; 20130243119; 20130243122; 20130243135; 20130257530; 20130271212; 20130272367; 20130285742; 20130301487; 20130303103; 20130315291; 20130321078; 20130330082; 20130336377; 20140009224; 20140029658; 20140029660; 20140030995; 20140031651; 20140036969; 20140044318; 20140044319; 20140044320; 20140044321; 20140072074; 20140077981; 20140081157; 20140086356; 20140086361; 20140095129; 20140107832; 20140126670; 20140126675; 20140133848; 20140140250; 20140161207; 20140167704; 20140172338; 20140198959; 20140213919; 20140225451; 20140226828; 20140229132; 20140247906; 20140266431; 20140269857; 20140269970; 20140269989; 20140269990; 20140270405; 20140278303; 20140279778; 20140292406; 20140292412; 20140294119; 20140294252; 20140313946; 20140314176; 20140314181; 20140314182; 20140317163; 20140323891; 20140333376; 20140372091; 20150003625; 20150005902; 20150016567; 20150018632; 20150025328; 20150031317; 20150031969; 20150032788; 20150043678; 20150051513; 20150061911; 20150070089; 20150077180; 20150078484; 20150092830; 20150098710; 20150104196; 20150131757; 20150156003; 20150156004; 20150162881; 20150172081; 20150180495; 20150193565; 20150193666; 20150194989; 20150202440; 20150214987; 20150215937; 20150223748; 20150241996; 20150249889; 20150256216; 20150270856; 20150270865; 20150288375; 20150295643; 20150311927; 20150311973; 20150311985; 20150322647; 20150326190; 20150333781; 20150357975; 20150358042; 20150358191; 20150381216; 20150381220; 20150381821; 20160005419; 20160022161; 20160028433; 20160034421; 20160036472; 20160036528; 20160065311; 20160079933; 20160087604; 20160087657; 20160099776; 20160111110; 20160117430; 20160124903; 20160126903; 20160127113; 20160132735; 20160134380; 20160156375; 20160162042; 20160173117; 20160191020; 20160218752; 20160225385; 20160241277; 20160248531; 20160259960; 20160261241; 20160269210; 20160287871; 20160308619; 20160309042; 20160316283; 20160329927; 20160334466; 20160336762; 20160352427; 20160359552; 20160373212; 20160380661; 20160380700; 20170012585; 20170012709; 20170014032; 20170032184; 20170033809; 20170041124; 20170043166; 20170047899; 20170061045; 20170063312; 20170077944; 20170077945; 20170078023; 20170078027; 20170093497; 20170095195; 20170104503; 20170108943; 20170117854; 20170141807; 20170141938; 20170163465; 20170170999; 20170180061; 20170195053; 20170207934; 20170214468; 20170214470; 20170222717; 20170244582; 20170245054; 20170245079; 20170255593; 20170272283; 20170304625; 20170322243; 20170324421; 20170338841; 20170338842; 20170339569; 20170346510; 20170366209; 20170366259; 20170373647; 20170373759; 20180013456; 20180013495; 20180026586; 20180026673; 20180041219; 20180062674; 20180070394; 20180102850; 20180131502; 20180167042; 20180167092; 20180167093; 20180180420; 20180191448; 20180219566; 20180254769; 20180262370; 20180269988; 20180279197; 20180294879; 20180294884; 20180302111; 20180309465; 20180316320; 20180331814; 20180333580; 20180367219; 20190007075; 20190013867; 20190013874; 20190013991; 20190020415; 20190028131; 20190030334; 20190036622; and 20190042536.

5G References

U.S. Pat. Nos. 6,675,125; 6,778,966; 7,027,981; 7,190,292; 7,206,420; 7,212,640; 7,558,391; 7,865,177; 8,085,943; 8,599,014; 8,725,706; 8,776,625; 8,898,567; 8,989,762; 9,160,579; 9,203,654; 9,235,268; 9,401,823; 9,432,564; 9,460,617; 9,531,427; 9,544,006; 9,564,927; 9,565,045; 9,613,408; 9,621,387; 9,660,851; 9,680,670; 9,686,112; 9,712,238; 9,712,350; 9,712,354; 9,713,019; 9,722,318; 9,729,281; 9,729,378; 9,742,521; 9,749,083; 9,774,476; 9,859,981; 9,871,679; 9,876,530; 9,877,206; 9,882,608; 9,893,919; 9,899,182; 9,900,048; 9,900,122; 9,900,123; 9,900,190; 9,912,436; 9,929,755; 9,942,074; 9,998,172; 9,998,187; 10,003,364; 10,027,397; 10,027,427; 10,027,523; 10,033,107; 10,033,108; 10,050,815; 10,051,483; 10,051,488; 10,062,970; 10,063,354; 10,069,467; 10,069,535; 10,079,652; 10,084,562; 10,090,594; 10,096,883; 10,103,777; 10,123,217; 10,129,057; 10,135,145; 10,148,016; 10,148,360; 10,168,501; 10,170,840; 10,171,158; 10,191,376; 10,198,582; 10,200,106; 10,205,212; 10,205,231; 10,205,482; 10,205,655; 10,211,855; 10,212,014; 10,218,405; 10,224,634; 20020051546; 20020085725; 20020103619; 20020172374; 20020172376; 20020172378; 20030035549; 20030055635; 20030098805; 20030112088; 20090221257; 20110238690; 20110249024; 20110252320; 20110288457; 20120112908; 20130110974; 20130201316; 20140226035; 20150146805; 20150146806; 20150230105; 20150280945; 20150310739; 20160093029; 20160149665; 20160149731; 20160197642; 20160218891; 20160226681; 20160352361; 20160352362; 20160352419; 20170012862; 20170018851; 20170018852; 20170019131; 20170026095; 20170032129; 20170033465; 20170033466; 20170033953; 20170033954; 20170063430; 20170078400; 20170085003; 20170085336; 20170093693; 20170104617; 20170110795; 20170110804; 20170111805; 20170134205; 20170201288; 20170229782; 20170230083; 20170245157; 20170269481; 20170271117; 20170288917; 20170295048; 20170311307; 20170317781; 20170317782; 20170317783; 20170317858; 20170318482; 20170331899; 20180013452; 20180034912; 20180048497; 20180054232; 20180054233; 20180054234; 20180054268; 20180062886; 20180069594; 20180069731; 20180076947; 20180076979; 20180076982; 20180076988; 20180091195; 20180115040; 20180115058; 20180123256; 20180123257; 20180123749; 20180123836; 20180123856; 20180123897; 20180124181; 20180131406; 20180131541; 20180145411; 20180145412; 20180145414; 20180145415; 20180151957; 20180152262; 20180152330; 20180152925; 20180159195; 20180159196; 20180159197; 20180159228; 20180159229; 20180159230; 20180159232; 20180159240; 20180159243; 20180159615; 20180166761; 20180166784; 20180166785; 20180166787; 20180167105; 20180167148; 20180175892; 20180175978; 20180198668; 20180205399; 20180205481; 20180227158; 20180248592; 20180254754; 20180254924; 20180262243; 20180278693; 20180278694; 20180294897; 20180301812; 20180302145; 20180309206; 20180323826; 20180324005; 20180324006; 20180324021; 20180324601; 20180331413; 20180331720; 20180331721; 20180331871; 20180343304; 20180351687; 20180358678; 20180359126; 20180375940; 20190013577; 20190013837; 20190013838; 20190020530; 20190036222; 20190052505; 20190074563; 20190074564; 20190074565; 20190074568; 20190074580; 20190074584; 20190074597; 20190074598; 20190074864; 20190074865; and 20190074878.

Claims

1. A distortion-compensating processor, comprising: at least one automated processor configured to decompose a non-linearly distorted signal derived from an information signal, received from a channel having a channel non-linear distortion into a truncated series expansion of at least third order with memory comprising a series of terms, each term representing incremental non-linearity order and associated delay;an adaptive multi-layer feedforward deep neural network comprising a plurality of hidden layers, and at least one dropout layer, receiving as inputs the series of terms, and producing an equalized output signal; andan output port configured to present the equalized output signal,the multi-layer feedforward deep neural network being trained with respect to the channel non-linear distortion associated with communication of a series of symbols using training data comprising the series of terms, to equalize the signal,the multi-layer feedforward deep neural network being configured to receive the respective terms associated with incremental non-linearity orders and associated delay values, and to selectively produce the equalized output signal, representing the information signal wherein the channel non-linear distortion is reduced.
2. The distortion-compensating processor according to claim 1, wherein the training data comprises a set of small amplitude training signals to estimate a channel response and a set of large amplitude training signals to estimate a power amplifier non-linearity.
3. The distortion-compensating processor according to claim 1, wherein the information signal is distorted by amplification by a radio frequency power amplifier and transmission through a radio frequency communication channel, wherein the non-linearly distorted signal is received by the at least one automated processor from a radio receiver.
4. The distortion-compensating processor according to claim 1, wherein the series expansion of at least third order with memory comprises a Volterra series expansion.
5. The distortion-compensating processor according to claim 4, wherein the terms of the Volterra series expansion are defined by:
6. The distortion-compensating processor according to claim 1, wherein: x(n) is a signal sequence representing information in the non-linearly distorted signal y(n) distorted by an analog process within a channel h;
7. The distortion-compensating processor according to claim 6, wherein:
8. The distortion-compensating processor according to claim 1, wherein the truncated series expansion of at least third order with memory comprises at least fifth order terms, and the deep multi-layer feedforward neural network has at least two convolutional network layers.
9. The distortion-compensating processor according to claim 1, wherein the deep multi-layer feedforward neural network comprises at least three hidden layers, each hidden layer comprising at least 10 feature maps, and a fully connected layer subsequent to the at least three hidden layers.
10. The distortion-compensating processor according to claim 1, wherein the non-linearly distorted signal comprises a frequency division multiplexed radio frequency modulated set of signals representing the information signal distorted by a radio frequency power amplifier;further comprising a frequency division multiplexed radio frequency signal demodulator configured to demodulate the equalized output signal as the set of symbols.
11. A method of compensating for a distortion, comprising: decomposing a non-linearly distorted signal received from a channel having a channel non-linear distortion into a truncated series expansion of at least third order with memory, based on an information signal communicated through the channel, the decomposition comprising a series of terms, each term representing incremental non-linearity order and associated delay, using at least one automated processor;equalizing the non-linearly distorted signal with an automated equalizer comprising a multi-layer feedforward deep neural network comprising a plurality of hidden layers and at least one dropout layer, by receiving coefficients of respective terms associated with respective incremental non-linearity orders and associated delay by the automated equalizer, and producing selectively from the equalizer an output signal representing the information signal wherein the channel non-linear distortion is reduced; andupdating the multi-layer feedforward deep neural network to reduce error of the output signal with respect to the information signal.
12. The method according to claim 11, wherein the non-linearly distorted signal comprises a frequency division multiplexed signal amplified and distorted by a radio frequency power amplifier, received though a radio frequency receiver, further comprising demodulating information contained within the frequency division multiplexed signal from the output signal.
13. The method according to claim 11, further comprising outputting the output signal, wherein the truncated series expansion of at least third order with memory comprises at least fifth order terms and the multi-layer feedforward deep neural network comprises at least two hidden layers.
14. The method according to claim 11, wherein the series expansion of at least third order with memory comprises a Volterra series expansion having terms defined by:
15. The method according to claim 11, wherein: x(n) is a signal sequence representing information in the signal y(n) distorted by an analog process within a channel h;
16. The method according to claim 11, wherein the multi-layer feedforward deep neural network comprises at least one convolutional layer, followed by a fully connected layer with dropout comprising the at least one dropout layer, and an output layer.
17. The method according to claim 11, wherein the multi-layer feedforward deep neural network comprises at least three one-dimensional convolutional layers each comprising at least 10 feature maps, followed by a first fully-connected layer with the at least one dropout layer provided for regularization, and a second fully-connected layer which produces the output with respect to a delay, the multi-layer feedforward deep neural network producing an output tensor having two dimensions,wherein the at three one-dimensional convolutional layers and the first fully connected layer use a sigmoid activation function, and the fully connected output layer uses a linear activation function.
18. A non-linear distortion-compensating processor, comprising: an input signal processor configured to decompose a received non-linearly distorted signal based on an information signal communicated through a channel, into a truncated Volterra series expansion, the truncated Volterra series expansion comprising a series of terms, each term comprising a sum of multidimensional convolutions of at least third order each with an associated time delay component; anda multi-layer feedforward deep neural network comprising a plurality of hidden neural network layers comprising at least one convolutional neural network layer and at least one dropout layer, trained with respect to a non-linear distortion of the information signal represented in the non-linearly distorted signal, and updated dependent on equalizer error, to receive the series of terms of the truncated Volterra series expansion, and to selectively produce an output signal representing the non-linearly distorted signal at least partially compensated for the non-linear distortion.
19. The non-linear distortion-compensating processor according to claim 18, further comprising a demodulator configured to demodulate information modulated in the non-linearly distorted signal, wherein the non-linearly distorted signal comprises a frequency division multiplexed signal distorted by a power amplifier and a communication channel.
20. The non-linear distortion-compensating processor according to claim 18, wherein the multi-layer feedforward deep neural network comprises a plurality of convolutional layers, each convolutional layer comprising at least 10 feature maps, followed by a first fully connected layer with the at least one dropout layer provided for regularization, and a second fully connected layer for output, wherein the plurality of convolutional layers and the first fully connected layer use a non-linear activation function, and the second fully connected layer uses a linear activation function.
21. A method of compensating for a distortion, comprising: decomposing a signal received from a channel having a channel non-linear distortion into a truncated series expansion of at least third order with memory, using at least one automated processor; andequalizing the signal with an automated equalizer comprising a multi-layer feedforward deep neural network trained having at least one convolutional layer, followed by a fully connected layer with dropout, and an output layer, with respect to the channel non-linear distortion, by receiving terms of the truncated series expansion of at least third order with memory by the automated equalizer, and producing selectively from the equalizer an output representing the signal wherein the channel non-linear distortion is reduced.

CROSS REFERENCE TO RELATED APPLICATIONS

The present application is a Continuation of U.S. patent application Ser. No. 17/234,102, filed Apr. 19, 2021, now U.S. Pat. No. 11,451,419, issued Sep. 20, 2022, which is a Continuation of U.S. patent application Ser. No. 16/812,229, filed Mar. 6, 2020, now U.S. Pat. No. 10,985,951, issued Apr. 20, 2021, which Claims benefit of priority under 35 U.S.C. § 119(e) from, and is a non-provisional of, U.S. Provisional Patent Application No. 62/819,054, filed Mar. 15, 2019, the entirety of which is expressly incorporated herein by reference.

US Referenced Citations (1217)

Number	Name	Date	Kind
4615038	Lim et al.	Sep 1986	A
4669116	Agazzi et al.	May 1987	A
4870371	Gottwald et al.	Sep 1989	A
5038187	Zhou	Aug 1991	A
5309481	Viviano et al.	May 1994	A
5329586	Agazzi	Jul 1994	A
5424680	Nazarathy et al.	Jun 1995	A
5438625	Klippel	Aug 1995	A
5539774	Nobakht et al.	Jul 1996	A
5647023	Agazzi et al.	Jul 1997	A
5692011	Nobakht et al.	Nov 1997	A
5694476	Klippel	Dec 1997	A
5744969	Grochowski et al.	Apr 1998	A
5745597	Agazzi et al.	Apr 1998	A
5790692	Price et al.	Aug 1998	A
5792062	Poon et al.	Aug 1998	A
5815585	Klippel	Sep 1998	A
5889823	Agazzi et al.	Mar 1999	A
5924086	Mathur et al.	Jul 1999	A
5938594	Poon et al.	Aug 1999	A
5991023	Morawski et al.	Nov 1999	A
6002479	Barwicz et al.	Dec 1999	A
6005952	Klippel	Dec 1999	A
6064265	Yun et al.	May 2000	A
6166599	Aparin et al.	Dec 2000	A
6181754	Chen	Jan 2001	B1
6201455	Kusunoki	Mar 2001	B1
6201839	Kavcic et al.	Mar 2001	B1
6236837	Midya	May 2001	B1
6240278	Midya et al.	May 2001	B1
6288610	Miyashita	Sep 2001	B1
6335767	Twitchell et al.	Jan 2002	B1
6351740	Rabinowitz	Feb 2002	B1
6381212	Larkin	Apr 2002	B1
6393259	Kusunoki	May 2002	B1
6406438	Thornton	Jun 2002	B1
6408079	Katayama et al.	Jun 2002	B1
6438180	Kavcic et al.	Aug 2002	B1
6453308	Zhao et al.	Sep 2002	B1
6504885	Chen	Jan 2003	B1
6510257	Barwicz et al.	Jan 2003	B1
6512417	Booth et al.	Jan 2003	B2
6532272	Ryan et al.	Mar 2003	B1
6563870	Schenk	May 2003	B1
6600794	Agarossi et al.	Jul 2003	B1
6633208	Salkola et al.	Oct 2003	B2
6636115	Kim et al.	Oct 2003	B2
6668256	Lynch	Dec 2003	B1
6675125	Bizjak	Jan 2004	B2
6687235	Chu	Feb 2004	B1
6690693	Crowder	Feb 2004	B1
6697768	Jones et al.	Feb 2004	B2
6711094	Katz et al.	Mar 2004	B1
6714481	Katz et al.	Mar 2004	B1
6718087	Choa	Apr 2004	B2
6775646	Tufillaro et al.	Aug 2004	B1
6778966	Bizjak	Aug 2004	B2
6788719	Crowder	Sep 2004	B2
6812792	Mattsson et al.	Nov 2004	B2
6826331	Barwicz et al.	Nov 2004	B2
6839657	Verbeyst et al.	Jan 2005	B2
6850871	Barford et al.	Feb 2005	B1
6868380	Kroeker	Mar 2005	B2
6885954	Jones et al.	Apr 2005	B2
6895262	Cortes et al.	May 2005	B2
6922552	Noori	Jul 2005	B2
6934655	Jones et al.	Aug 2005	B2
6940790	Powelson et al.	Sep 2005	B1
6947857	Jones et al.	Sep 2005	B2
6951540	Ebbini et al.	Oct 2005	B2
6954476	Coldren et al.	Oct 2005	B2
6956433	Kim et al.	Oct 2005	B2
6982939	Powelson et al.	Jan 2006	B2
6992519	Vilander et al.	Jan 2006	B2
6999201	Shimizu	Feb 2006	B1
6999510	Batruni	Feb 2006	B2
7007253	Gullapalli et al.	Feb 2006	B2
7016823	Yang	Mar 2006	B2
7027981	Bizjak	Apr 2006	B2
7061943	Coldren et al.	Jun 2006	B2
7065511	Zhao et al.	Jun 2006	B2
7071797	Ye	Jul 2006	B2
7084974	Barwicz et al.	Aug 2006	B1
7092043	Vorenkamp et al.	Aug 2006	B2
7113037	Nezami	Sep 2006	B2
7123663	De Gaudenzi et al.	Oct 2006	B2
7151405	Nezami	Dec 2006	B2
7176757	Nakatani	Feb 2007	B2
7190292	Bizjak	Mar 2007	B2
7206420	Bizjak	Apr 2007	B2
7209566	Griniasty	Apr 2007	B2
7212640	Bizjak	May 2007	B2
7212933	Kouri et al.	May 2007	B2
7236156	Liberty et al.	Jun 2007	B2
7236212	Carr et al.	Jun 2007	B2
7239301	Liberty et al.	Jul 2007	B2
7239668	De Gaudenzi et al.	Jul 2007	B2
7251297	Agazzi	Jul 2007	B2
7268620	Nygren et al.	Sep 2007	B2
7272594	Lynch et al.	Sep 2007	B1
7286009	Andersen et al.	Oct 2007	B2
7295961	Root et al.	Nov 2007	B2
7304591	Raphaeli	Dec 2007	B2
7305639	Floyd et al.	Dec 2007	B2
7308032	Capofreddi	Dec 2007	B2
7333559	Song et al.	Feb 2008	B2
7348844	Jaenecke	Mar 2008	B2
7400807	Minelly et al.	Jul 2008	B2
7403884	Hemmett	Jul 2008	B2
7412469	Dalipi	Aug 2008	B2
7423699	Vorenkamp et al.	Sep 2008	B2
7436883	Batruni	Oct 2008	B2
7443326	Raphaeli	Oct 2008	B2
7489298	Liberty et al.	Feb 2009	B2
7512900	Lynch et al.	Mar 2009	B2
7542518	Kim et al.	Jun 2009	B2
7551668	Higashino et al.	Jun 2009	B2
7558391	Bizjak	Jul 2009	B2
7570856	Minelly et al.	Aug 2009	B1
7571401	Rao et al.	Aug 2009	B1
7576606	Andersen et al.	Aug 2009	B2
7589725	Snyder et al.	Sep 2009	B2
7590518	Phillips	Sep 2009	B2
7602240	Gao et al.	Oct 2009	B2
7606539	Singerl et al.	Oct 2009	B2
7610183	Danko	Oct 2009	B2
7657405	Singerl et al.	Feb 2010	B2
7720232	Oxford	May 2010	B2
7720236	Oxford	May 2010	B2
7728658	Andersen et al.	Jun 2010	B2
7729446	Copeland	Jun 2010	B2
7733177	Borkar et al.	Jun 2010	B1
7746955	Rexberg	Jun 2010	B2
7755425	Klingberg et al.	Jul 2010	B2
7760887	Oxford	Jul 2010	B2
7773692	Copeland et al.	Aug 2010	B2
7774176	Rao et al.	Aug 2010	B2
7795858	Tufillaro et al.	Sep 2010	B2
7796960	Rashev et al.	Sep 2010	B1
7808315	Goodman et al.	Oct 2010	B2
7812666	Chieng et al.	Oct 2010	B2
7821337	Yamanouchi et al.	Oct 2010	B2
7821581	Vorenkamp et al.	Oct 2010	B2
7822146	Copeland	Oct 2010	B2
7826624	Oxford	Nov 2010	B2
7847631	Jiang et al.	Dec 2010	B2
7852913	Agazzi et al.	Dec 2010	B2
7853443	Hemmett	Dec 2010	B2
7864881	Hori et al.	Jan 2011	B2
7865177	Sorrells et al.	Jan 2011	B2
7873172	Lashkari	Jan 2011	B2
7885025	Eppler et al.	Feb 2011	B2
7885797	Koppl et al.	Feb 2011	B2
7889007	Kim et al.	Feb 2011	B2
7894788	Keehr et al.	Feb 2011	B2
7895006	Thompson	Feb 2011	B2
7899416	McCallister et al.	Mar 2011	B2
7902925	Kim et al.	Mar 2011	B2
7903137	Oxford et al.	Mar 2011	B2
7924942	Rexberg	Apr 2011	B2
7929375	Nuttall et al.	Apr 2011	B2
7932782	Lau et al.	Apr 2011	B2
7970150	Oxford	Jun 2011	B2
7970151	Oxford et al.	Jun 2011	B2
7979837	Rao et al.	Jul 2011	B1
7991073	Utsunomiya et al.	Aug 2011	B2
7991167	Oxford	Aug 2011	B2
7995674	Hori et al.	Aug 2011	B2
8005858	Lynch et al.	Aug 2011	B1
8023668	Pfaffinger	Sep 2011	B2
8031882	Ding	Oct 2011	B2
8039871	Nogami	Oct 2011	B2
8045066	Vorenkamp et al.	Oct 2011	B2
8046199	Copeland	Oct 2011	B2
8065060	Danko	Nov 2011	B2
8085943	Bizjak	Dec 2011	B2
8089689	Savage-Leuchs	Jan 2012	B1
8105270	Hunter	Jan 2012	B2
8139630	Agazzi et al.	Mar 2012	B2
8148983	Biber et al.	Apr 2012	B2
8149950	Kim et al.	Apr 2012	B2
8160191	Row et al.	Apr 2012	B2
8165854	Wei	Apr 2012	B1
8170508	Davies	May 2012	B2
8185853	Kim et al.	May 2012	B2
8193566	Nogami	Jun 2012	B2
8195103	Waheed et al.	Jun 2012	B2
8199399	Savage-Leuchs	Jun 2012	B1
8213880	van Zelm et al.	Jul 2012	B2
8244787	Principe et al.	Aug 2012	B2
8260732	Al-Duwaish et al.	Sep 2012	B2
8265583	Venkataraman	Sep 2012	B1
8270530	Hamada et al.	Sep 2012	B2
8294605	Pagnanelli	Oct 2012	B1
8295790	Koren et al.	Oct 2012	B2
8306488	Rashev et al.	Nov 2012	B2
8310312	Lee et al.	Nov 2012	B2
8315970	Zalay et al.	Nov 2012	B2
8331511	Beidas et al.	Dec 2012	B2
8331879	van Zelm et al.	Dec 2012	B2
8345348	Savage-Leuchs	Jan 2013	B1
8346692	Rouat et al.	Jan 2013	B2
8346693	Al-Duwaish et al.	Jan 2013	B2
8346711	Al-Duwaish et al.	Jan 2013	B2
8346712	Rizvi et al.	Jan 2013	B2
8351876	McCallister et al.	Jan 2013	B2
8354884	Braithwaite	Jan 2013	B2
8355684	Yu et al.	Jan 2013	B2
8358169	Sen et al.	Jan 2013	B2
8364095	van Zelm et al.	Jan 2013	B2
8369447	Fuller et al.	Feb 2013	B2
8369595	Derakhshani et al.	Feb 2013	B1
8380773	Batruni	Feb 2013	B2
8390375	Miyashita	Mar 2013	B2
8390376	Bai	Mar 2013	B2
8396693	Danko	Mar 2013	B2
8410843	Goodman et al.	Apr 2013	B2
8410850	Mazzucco et al.	Apr 2013	B2
8412133	Davies	Apr 2013	B2
8421534	Kim et al.	Apr 2013	B2
8432220	Peyresoubes et al.	Apr 2013	B2
8437513	Derakhshani et al.	May 2013	B1
8463582	Song et al.	Jun 2013	B2
8467438	Beidas	Jun 2013	B2
8477581	Liu et al.	Jul 2013	B1
8483343	Agazzi et al.	Jul 2013	B2
8483450	Derakhshani et al.	Jul 2013	B1
8487706	Li et al.	Jul 2013	B2
8489047	McCallister et al.	Jul 2013	B2
8494463	Davies	Jul 2013	B2
8498369	Forrester et al.	Jul 2013	B2
8509347	Kim et al.	Aug 2013	B2
8509712	van Zelm et al.	Aug 2013	B2
8519440	Nogami	Aug 2013	B2
8532215	Huang et al.	Sep 2013	B2
8532964	Wei	Sep 2013	B2
8538039	Pfaffinger	Sep 2013	B2
8564368	Bai	Oct 2013	B1
8565343	Husted et al.	Oct 2013	B1
8577311	Wolf et al.	Nov 2013	B2
8587375	Kim et al.	Nov 2013	B2
8599014	Prykari et al.	Dec 2013	B2
8599050	Rachid et al.	Dec 2013	B2
8605814	McCallister et al.	Dec 2013	B2
8605819	Lozhkin	Dec 2013	B2
8611190	Hughes et al.	Dec 2013	B1
8611459	McCallister	Dec 2013	B2
8611820	Gilmore	Dec 2013	B2
8615208	McCallister et al.	Dec 2013	B2
8619905	Ishikawa et al.	Dec 2013	B2
8620631	Al-Duwaish	Dec 2013	B2
8626089	Singerl et al.	Jan 2014	B2
8649743	McCallister et al.	Feb 2014	B2
8675925	Derakhshani et al.	Mar 2014	B2
8704595	Kim et al.	Apr 2014	B2
8705166	Savage-Leuchs	Apr 2014	B1
8712345	Ishikawa et al.	Apr 2014	B2
8718178	Carbone et al.	May 2014	B1
8718209	Lozhkin	May 2014	B2
8724857	Derakhshani et al.	May 2014	B2
8725706	Arrasvuori et al.	May 2014	B2
8737937	Ishikawa et al.	May 2014	B2
8737938	Rashev et al.	May 2014	B2
8744141	Derakhshani et al.	Jun 2014	B2
8744377	Rimini et al.	Jun 2014	B2
8758271	Hunter et al.	Jun 2014	B2
8761409	Pfaffinger	Jun 2014	B2
8766917	Liberty et al.	Jul 2014	B2
8767869	Rimini et al.	Jul 2014	B2
8776625	Questo et al.	Jul 2014	B2
8780693	Kim et al.	Jul 2014	B2
8787628	Derakhshani et al.	Jul 2014	B1
8798559	Kilambi et al.	Aug 2014	B2
8804807	Liu et al.	Aug 2014	B2
8804871	Rimini et al.	Aug 2014	B2
8811532	Bai	Aug 2014	B2
8823452	Chen et al.	Sep 2014	B2
8831074	Agazzi et al.	Sep 2014	B2
8831133	Azadet et al.	Sep 2014	B2
8831135	Utsunomiya et al.	Sep 2014	B2
8838218	Khair	Sep 2014	B2
8843088	van Zelm et al.	Sep 2014	B2
8843089	Davies	Sep 2014	B2
8849611	Haviland et al.	Sep 2014	B2
8855175	Wyville et al.	Oct 2014	B2
8855234	Kim et al.	Oct 2014	B2
8867601	Beidas	Oct 2014	B2
8874411	Ishikawa et al.	Oct 2014	B2
8885765	Zhang et al.	Nov 2014	B2
8886341	van Zelm et al.	Nov 2014	B1
8891701	Eliaz et al.	Nov 2014	B1
8896471	Pagnanelli	Nov 2014	B1
8897351	Kravtsov	Nov 2014	B2
8898567	Arrasvuori et al.	Nov 2014	B2
8903192	Malik et al.	Dec 2014	B2
8909176	van Zelm et al.	Dec 2014	B2
8909328	Chon	Dec 2014	B2
8933752	Nagatani et al.	Jan 2015	B2
8934573	McCallister et al.	Jan 2015	B2
8958470	Bolstad et al.	Feb 2015	B2
8964901	Kim et al.	Feb 2015	B2
8964996	Klippel et al.	Feb 2015	B2
8971834	Keehr et al.	Mar 2015	B2
8976896	McCallister et al.	Mar 2015	B2
8989762	Negus et al.	Mar 2015	B1
8994657	Liberty et al.	Mar 2015	B2
8995571	Chen	Mar 2015	B2
8995835	Yan et al.	Mar 2015	B2
9008153	Pang et al.	Apr 2015	B2
9014299	Teterwak	Apr 2015	B2
9019643	Medard et al.	Apr 2015	B2
9020454	Waheed et al.	Apr 2015	B2
9025607	Zeger et al.	May 2015	B2
9031168	Liu	May 2015	B1
9036734	Mauer et al.	May 2015	B1
9048865	Pagnanelli	Jun 2015	B2
9048900	Pratt et al.	Jun 2015	B2
9071313	Monsen	Jun 2015	B2
9077508	Koike-Akino et al.	Jul 2015	B2
9088472	Jain et al.	Jul 2015	B1
9094036	Rachid et al.	Jul 2015	B2
9094151	Alonso et al.	Jul 2015	B2
9104921	Derakhshani et al.	Aug 2015	B2
9106304	Row et al.	Aug 2015	B2
9130628	Mittal et al.	Sep 2015	B1
9137082	Ali	Sep 2015	B1
9137492	Lima et al.	Sep 2015	B2
9143274	Zeger et al.	Sep 2015	B2
9160280	Abdelhafiz et al.	Oct 2015	B1
9160310	Velazquez et al.	Oct 2015	B2
9160579	Tan et al.	Oct 2015	B1
9160687	Haeupler et al.	Oct 2015	B2
9166610	Klippel	Oct 2015	B2
9166635	Carbone et al.	Oct 2015	B2
9166698	Bae et al.	Oct 2015	B2
9171534	Tronchin et al.	Oct 2015	B2
9184784	Ding et al.	Nov 2015	B2
9185529	Medard et al.	Nov 2015	B2
9189458	Langer et al.	Nov 2015	B1
9191041	Mkadem et al.	Nov 2015	B2
9191049	Mikhemar et al.	Nov 2015	B2
9199860	MacArthur	Dec 2015	B2
9203654	Terry	Dec 2015	B2
9209753	Xiao et al.	Dec 2015	B2
9209841	Yu et al.	Dec 2015	B2
9214968	Wang	Dec 2015	B2
9214969	Hammi	Dec 2015	B2
9225295	Ishikawa et al.	Dec 2015	B2
9225501	Azadet	Dec 2015	B2
9231530	Kaushik et al.	Jan 2016	B1
9231647	Polydoros et al.	Jan 2016	B2
9231801	Rimini et al.	Jan 2016	B2
9235268	Arrasvuori et al.	Jan 2016	B2
9236996	Khandani	Jan 2016	B2
9246525	Breynaert et al.	Jan 2016	B2
9246731	Kim et al.	Jan 2016	B2
9252798	Rachid et al.	Feb 2016	B2
9252821	Shor et al.	Feb 2016	B2
9253608	Medard et al.	Feb 2016	B2
9257943	Onishi	Feb 2016	B2
9258156	Wloczysiak	Feb 2016	B2
9261978	Liberty et al.	Feb 2016	B2
9264153	Kim et al.	Feb 2016	B2
9265461	Hunter et al.	Feb 2016	B2
9270304	Monsen	Feb 2016	B2
9270512	Eliaz et al.	Feb 2016	B2
9271123	Medard et al.	Feb 2016	B2
9276602	Pagnanelli	Mar 2016	B1
9294113	Feizi-Khankandi et al.	Mar 2016	B2
9304501	Danko	Apr 2016	B2
9306606	Zhang	Apr 2016	B2
9311535	Derakhshani et al.	Apr 2016	B2
9312892	Chang	Apr 2016	B2
9314623	Bardakjian et al.	Apr 2016	B2
9322906	Seppa et al.	Apr 2016	B2
9337781	Hammi	May 2016	B2
9337783	Matsubara et al.	May 2016	B2
9352155	Moehlis et al.	May 2016	B2
9361681	Derakhshani et al.	Jun 2016	B2
9361936	Medard et al.	Jun 2016	B2
9362869	Dechen et al.	Jun 2016	B2
9362942	Hammler	Jun 2016	B1
9363068	Azadet et al.	Jun 2016	B2
9369093	Gustavsson	Jun 2016	B2
9369255	Medard et al.	Jun 2016	B2
9369541	Medard et al.	Jun 2016	B2
9397516	Hunter et al.	Jul 2016	B2
9401823	Terry	Jul 2016	B2
9404950	Lafontaine et al.	Aug 2016	B2
9413516	Khandani	Aug 2016	B2
9419722	Winzer et al.	Aug 2016	B2
9431972	Xu	Aug 2016	B1
9432564	Nurmenniemi	Aug 2016	B2
9438178	Sulimarski et al.	Sep 2016	B1
9438356	Kim et al.	Sep 2016	B2
9439597	Warrick et al.	Sep 2016	B2
9451920	Khair	Sep 2016	B2
9460246	Williams	Oct 2016	B2
9460617	Beaurepaire et al.	Oct 2016	B2
9461597	Abdelrahman et al.	Oct 2016	B2
9461676	Santucci et al.	Oct 2016	B2
9473077	Feldman et al.	Oct 2016	B2
9479322	Khandani	Oct 2016	B2
9509331	Pagnanelli	Nov 2016	B1
9509350	Magesacher et al.	Nov 2016	B1
9517030	Hunter et al.	Dec 2016	B2
9531427	Henry et al.	Dec 2016	B2
9531475	Agazzi et al.	Dec 2016	B2
9536539	Chang et al.	Jan 2017	B2
9537759	Calmon et al.	Jan 2017	B2
9544006	Henry et al.	Jan 2017	B2
9544126	Zeger et al.	Jan 2017	B2
9559831	Zeger et al.	Jan 2017	B2
9564876	Kim et al.	Feb 2017	B2
9564927	Fonseka et al.	Feb 2017	B2
9565045	Terry	Feb 2017	B2
9571312	Brandt-Pearce et al.	Feb 2017	B2
9575570	Liberty et al.	Feb 2017	B2
9590664	Rexberg et al.	Mar 2017	B2
9590668	Kim et al.	Mar 2017	B1
9595920	Li et al.	Mar 2017	B2
9595982	Weber et al.	Mar 2017	B2
9607003	Medard et al.	Mar 2017	B2
9607628	Gautama	Mar 2017	B2
9608676	Chen	Mar 2017	B2
9608718	Monsen et al.	Mar 2017	B2
9613408	Micovic et al.	Apr 2017	B2
9614554	Beidas et al.	Apr 2017	B2
9621387	Magers	Apr 2017	B1
9628119	Gal et al.	Apr 2017	B2
9628120	Yu et al.	Apr 2017	B2
9646116	Liu et al.	May 2017	B2
9647717	Xiong et al.	May 2017	B2
9654211	Zhao et al.	May 2017	B2
9654216	Nakashima et al.	May 2017	B2
9659120	Fehri et al.	May 2017	B2
9660593	Abdelrahman et al.	May 2017	B2
9660730	Rope	May 2017	B1
9660851	Hadani et al.	May 2017	B2
9665510	Kaushik et al.	May 2017	B2
9667292	Xue et al.	May 2017	B2
9674368	Kechichian et al.	Jun 2017	B2
9680423	Chen	Jun 2017	B2
9680497	Pagnanelli	Jun 2017	B2
9680670	Henry et al.	Jun 2017	B2
9686112	Terry	Jun 2017	B2
9697845	Hammarqvist	Jul 2017	B2
9705477	Velazquez	Jul 2017	B2
9706296	Uhle et al.	Jul 2017	B2
9712179	Rachid et al.	Jul 2017	B2
9712233	Deng et al.	Jul 2017	B1
9712238	Ashrafi et al.	Jul 2017	B2
9712350	Henry et al.	Jul 2017	B2
9712354	Hadani et al.	Jul 2017	B2
9713010	Khandani	Jul 2017	B2
9713019	Negus et al.	Jul 2017	B2
9722318	Adriazola et al.	Aug 2017	B2
9722646	Matthews et al.	Aug 2017	B1
9722691	Colavolpe et al.	Aug 2017	B2
9726701	Lafontaine et al.	Aug 2017	B2
9727677	Fehri et al.	Aug 2017	B2
9729281	Hadani et al.	Aug 2017	B2
9729378	Ahmed	Aug 2017	B1
9735741	Pratt et al.	Aug 2017	B2
9735800	Pagnanelli	Aug 2017	B2
9735811	Dalipi et al.	Aug 2017	B2
9735876	Duthel	Aug 2017	B2
9737258	Poon et al.	Aug 2017	B2
9742521	Henry et al.	Aug 2017	B2
9742599	Iyer Seshadri et al.	Aug 2017	B2
9746506	Lafontaine et al.	Aug 2017	B2
9749083	Henry et al.	Aug 2017	B2
9749161	Gal et al.	Aug 2017	B1
9755691	Kim et al.	Sep 2017	B2
9762268	Yang et al.	Sep 2017	B2
9768891	Ishikawa et al.	Sep 2017	B2
9774476	Nammi et al.	Sep 2017	B2
9778902	Azadet et al.	Oct 2017	B2
9780869	Rope	Oct 2017	B2
9780881	Rope et al.	Oct 2017	B1
9787351	Martineau et al.	Oct 2017	B2
9787459	Azadet	Oct 2017	B2
9794000	Ling	Oct 2017	B2
9800437	Kingery et al.	Oct 2017	B2
9800734	Kechichian et al.	Oct 2017	B2
9820311	Khandani	Nov 2017	B2
9831899	Boghrat et al.	Nov 2017	B1
9837970	Hammi et al.	Dec 2017	B2
9843346	Mundarath et al.	Dec 2017	B1
9859845	Sarbishaei et al.	Jan 2018	B2
9859981	Ashrafi et al.	Jan 2018	B2
9866183	Magesacher et al.	Jan 2018	B2
9871679	Henry et al.	Jan 2018	B2
9876530	Negus et al.	Jan 2018	B2
9877206	Henry et al.	Jan 2018	B2
9877265	Kim et al.	Jan 2018	B2
9882608	Adriazola et al.	Jan 2018	B2
9882648	Agazzi et al.	Jan 2018	B2
9887862	Zhou et al.	Feb 2018	B2
9893919	Kim et al.	Feb 2018	B2
9899182	Borodovsky	Feb 2018	B2
9900048	Hadani et al.	Feb 2018	B2
9900088	Rope	Feb 2018	B2
9900122	Henry et al.	Feb 2018	B2
9900123	Henry et al.	Feb 2018	B2
9900190	Henry et al.	Feb 2018	B2
9912435	Kim et al.	Mar 2018	B2
9912436	Henry et al.	Mar 2018	B2
9913194	Kim et al.	Mar 2018	B2
9923524	Carbone et al.	Mar 2018	B2
9923640	Okabe et al.	Mar 2018	B2
9923714	Lima et al.	Mar 2018	B2
9928212	Agee	Mar 2018	B2
9929755	Henry et al.	Mar 2018	B2
9935590	Sulimarski et al.	Apr 2018	B2
9935645	Tangudu et al.	Apr 2018	B1
9935715	Duthel	Apr 2018	B2
9935761	Azadet	Apr 2018	B2
9940938	Dick et al.	Apr 2018	B2
9941963	Magri et al.	Apr 2018	B2
9942074	Luo et al.	Apr 2018	B1
9953656	Dick et al.	Apr 2018	B2
9954384	Hunter et al.	Apr 2018	B2
9960794	Weissman et al.	May 2018	B2
9960804	Kim et al.	May 2018	B2
9960900	Azadet	May 2018	B2
9971920	Derakhshani et al.	May 2018	B2
9973279	Shen	May 2018	B2
9974957	Choi et al.	May 2018	B2
9983243	Lafontaine et al.	May 2018	B2
9998172	Barzegar et al.	Jun 2018	B1
9998187	Ashrafi et al.	Jun 2018	B2
9998223	Zhao et al.	Jun 2018	B2
9998406	Haeupler et al.	Jun 2018	B2
9999780	Weyh et al.	Jun 2018	B2
11451419	Li	Sep 2022	B2
20010051871	Kroeker	Mar 2001	A1
20010036334	Choa	Nov 2001	A1
20020041210	Booth et al.	Apr 2002	A1
20020051546	Bizjak	May 2002	A1
20020060827	Agazzi	May 2002	A1
20020075918	Crowder	Jun 2002	A1
20020085725	Bizjak	Jul 2002	A1
20020103619	Bizjak	Aug 2002	A1
20020126604	Powelson et al.	Sep 2002	A1
20020146993	Persico et al.	Oct 2002	A1
20020161539	Jones et al.	Oct 2002	A1
20020161542	Jones et al.	Oct 2002	A1
20020169585	Jones et al.	Nov 2002	A1
20020172374	Bizjak	Nov 2002	A1
20020172376	Bizjak	Nov 2002	A1
20020172378	Bizjak	Nov 2002	A1
20020178133	Zhao et al.	Nov 2002	A1
20020181521	Crowder et al.	Dec 2002	A1
20020186874	Price et al.	Dec 2002	A1
20030035549	Bizjak et al.	Feb 2003	A1
20030046045	Pileggi et al.	Mar 2003	A1
20030055635	Bizjak	Mar 2003	A1
20030057963	Verbeyst et al.	Mar 2003	A1
20030063854	Barwicz et al.	Apr 2003	A1
20030071684	Noori	Apr 2003	A1
20030098805	Bizjak	May 2003	A1
20030112088	Bizjak	Jun 2003	A1
20030142832	Meerkoetter et al.	Jul 2003	A1
20030195706	Korenberg	Oct 2003	A1
20030223507	De Gaudenzi et al.	Dec 2003	A1
20040019443	Jones et al.	Jan 2004	A1
20040044489	Jones et al.	Mar 2004	A1
20040130394	Mattsson et al.	Jul 2004	A1
20040136423	Coldren et al.	Jul 2004	A1
20040155707	Kim et al.	Aug 2004	A1
20040179629	Song et al.	Sep 2004	A1
20040208242	Batruni	Oct 2004	A1
20040258176	Mattsson et al.	Dec 2004	A1
20050021266	Kouri et al.	Jan 2005	A1
20050021319	Li et al.	Jan 2005	A1
20050031117	Browning et al.	Feb 2005	A1
20050031131	Browning et al.	Feb 2005	A1
20050031132	Browning et al.	Feb 2005	A1
20050031133	Browning et al.	Feb 2005	A1
20050031134	Leske	Feb 2005	A1
20050031137	Browning et al.	Feb 2005	A1
20050031138	Browning et al.	Feb 2005	A1
20050031139	Browning et al.	Feb 2005	A1
20050031140	Browning	Feb 2005	A1
20050049838	Danko	Mar 2005	A1
20050100065	Coldren et al.	May 2005	A1
20050141637	Domokos	Jun 2005	A1
20050141659	Capofreddi	Jun 2005	A1
20050174167	Vilander et al.	Aug 2005	A1
20050177805	Lynch et al.	Aug 2005	A1
20050180526	Kim et al.	Aug 2005	A1
20050226316	Higashino et al.	Oct 2005	A1
20050237111	Nygren et al.	Oct 2005	A1
20050243061	Liberty et al.	Nov 2005	A1
20050253806	Liberty et al.	Nov 2005	A1
20050270094	Nakatani	Dec 2005	A1
20050271216	Lashkari	Dec 2005	A1
20050273188	Barwicz et al.	Dec 2005	A1
20060039498	de Figueiredo et al.	Feb 2006	A1
20060052988	Farahani et al.	Mar 2006	A1
20060083389	Oxford et al.	Apr 2006	A1
20060093128	Oxford	May 2006	A1
20060095236	Phillips	May 2006	A1
20060104395	Batruni	May 2006	A1
20060104451	Browning et al.	May 2006	A1
20060133536	Rexberg	Jun 2006	A1
20060209982	De Gaudenzi et al.	Sep 2006	A1
20060222128	Raphaeli	Oct 2006	A1
20060239443	Oxford et al.	Oct 2006	A1
20060256974	Oxford	Nov 2006	A1
20060262942	Oxford	Nov 2006	A1
20060262943	Oxford	Nov 2006	A1
20060264187	Singerl et al.	Nov 2006	A1
20060269074	Oxford	Nov 2006	A1
20060269080	Oxford et al.	Nov 2006	A1
20060274904	Lashkari	Dec 2006	A1
20070005326	Koppl et al.	Jan 2007	A1
20070018722	Jaenecke	Jan 2007	A1
20070030076	Kim et al.	Feb 2007	A1
20070033000	Farahani et al.	Feb 2007	A1
20070063770	Rexberg	Mar 2007	A1
20070080841	Raphaeli	Apr 2007	A1
20070133713	Dalipi	Jun 2007	A1
20070133719	Agazzi et al.	Jun 2007	A1
20070136018	Fernandez et al.	Jun 2007	A1
20070136045	Rao et al.	Jun 2007	A1
20070152750	Andersen et al.	Jul 2007	A1
20070160221	Pfaffinger	Jul 2007	A1
20070168100	Danko	Jul 2007	A1
20070190952	Waheed et al.	Aug 2007	A1
20070229154	Kim et al.	Oct 2007	A1
20070237260	Hori et al.	Oct 2007	A1
20070247425	Liberty et al.	Oct 2007	A1
20070252651	Gao et al.	Nov 2007	A1
20070252813	Liberty et al.	Nov 2007	A1
20070276610	Korenberg	Nov 2007	A1
20080001947	Snyder et al.	Jan 2008	A1
20080032642	Singerl et al.	Feb 2008	A1
20080129379	Copeland	Jun 2008	A1
20080130787	Copeland	Jun 2008	A1
20080130788	Copeland	Jun 2008	A1
20080130789	Copeland et al.	Jun 2008	A1
20080152037	Kim et al.	Jun 2008	A1
20080158154	Liberty et al.	Jul 2008	A1
20080158155	Liberty et al.	Jul 2008	A1
20080180178	Gao et al.	Jul 2008	A1
20080240325	Agazzi et al.	Oct 2008	A1
20080261541	Langer	Oct 2008	A1
20080283882	Nogami	Nov 2008	A1
20080285640	McCallister et al.	Nov 2008	A1
20080293372	Principe et al.	Nov 2008	A1
20090003134	Nuttall et al.	Jan 2009	A1
20090027117	Andersen et al.	Jan 2009	A1
20090027118	Andersen et al.	Jan 2009	A1
20090058521	Fernandez	Mar 2009	A1
20090067643	Ding	Mar 2009	A1
20090072901	Yamanouchi et al.	Mar 2009	A1
20090075610	Keehr et al.	Mar 2009	A1
20090094304	Batruni	Apr 2009	A1
20090146740	Lau et al.	Jun 2009	A1
20090153132	Tufillaro et al.	Jun 2009	A1
20090185613	Agazzi et al.	Jul 2009	A1
20090221257	Sorrells et al.	Sep 2009	A1
20090256632	Klingberg et al.	Oct 2009	A1
20090287624	Rouat et al.	Nov 2009	A1
20090289706	Goodman et al.	Nov 2009	A1
20090291650	Singerl et al.	Nov 2009	A1
20090302938	Andersen et al.	Dec 2009	A1
20090302940	Fuller et al.	Dec 2009	A1
20090318983	Armoundas et al.	Dec 2009	A1
20100007489	Misra et al.	Jan 2010	A1
20100033180	Biber et al.	Feb 2010	A1
20100060355	Hamada et al.	Mar 2010	A1
20100090762	van Zelm et al.	Apr 2010	A1
20100093290	van Zelm et al.	Apr 2010	A1
20100094603	Danko	Apr 2010	A1
20100097714	Eppler et al.	Apr 2010	A1
20100114813	Zalay et al.	May 2010	A1
20100135449	Row et al.	Jun 2010	A1
20100148865	Borkar et al.	Jun 2010	A1
20100152547	Sterling et al.	Jun 2010	A1
20100156530	Utsunomiya et al.	Jun 2010	A1
20100183106	Beidas et al.	Jul 2010	A1
20100194474	Ishikawa et al.	Aug 2010	A1
20100199237	Kim et al.	Aug 2010	A1
20100254450	Narroschke et al.	Oct 2010	A1
20100283540	Davies	Nov 2010	A1
20100292602	Worrell et al.	Nov 2010	A1
20100292752	Bardakjian et al.	Nov 2010	A1
20100311361	Rashev et al.	Dec 2010	A1
20100312495	Haviland et al.	Dec 2010	A1
20110003570	Yu et al.	Jan 2011	A1
20110025414	Wolf et al.	Feb 2011	A1
20110028859	Chian	Feb 2011	A1
20110037518	Lee et al.	Feb 2011	A1
20110054354	Hunter et al.	Mar 2011	A1
20110054355	Hunter et al.	Mar 2011	A1
20110064171	Huang et al.	Mar 2011	A1
20110069749	Forrester et al.	Mar 2011	A1
20110081152	Agazzi et al.	Apr 2011	A1
20110085678	Pfaffinger	Apr 2011	A1
20110087341	Pfaffinger	Apr 2011	A1
20110096865	Hori et al.	Apr 2011	A1
20110102080	Chatterjee et al.	May 2011	A1
20110103455	Forrester et al.	May 2011	A1
20110110473	Keehr et al.	May 2011	A1
20110121897	Braithwaite	May 2011	A1
20110125684	Al-Duwaish et al.	May 2011	A1
20110125685	Rizvi et al.	May 2011	A1
20110125686	Al-Duwaish et al.	May 2011	A1
20110125687	Al-Duwaish et al.	May 2011	A1
20110140779	Koren et al.	Jun 2011	A1
20110144961	Ishikawa et al.	Jun 2011	A1
20110149714	Rimini et al.	Jun 2011	A1
20110177956	Korenberg	Jul 2011	A1
20110181360	Li et al.	Jul 2011	A1
20110204975	Miyashita	Aug 2011	A1
20110211842	Agazzi et al.	Sep 2011	A1
20110238690	Arrasvuori et al.	Sep 2011	A1
20110249024	Arrasvuori et al.	Oct 2011	A1
20110252320	Arrasvuori et al.	Oct 2011	A1
20110268226	Lozhkin	Nov 2011	A1
20110270590	Aparin	Nov 2011	A1
20110288457	Questo et al.	Nov 2011	A1
20110293051	Lozhkin	Dec 2011	A1
20120007153	Nogami	Jan 2012	A1
20120007672	Peyresoubes et al.	Jan 2012	A1
20120027070	Beidas	Feb 2012	A1
20120029663	Danko	Feb 2012	A1
20120086504	Tsukamoto	Apr 2012	A1
20120086507	Kim et al.	Apr 2012	A1
20120093376	Malik et al.	Apr 2012	A1
20120098481	Hunter et al.	Apr 2012	A1
20120098596	Nagatani et al.	Apr 2012	A1
20120112908	Prykari et al.	May 2012	A1
20120119810	Bai	May 2012	A1
20120140860	Rimini et al.	Jun 2012	A1
20120147993	Kim et al.	Jun 2012	A1
20120154040	Kim et al.	Jun 2012	A1
20120154041	Kim et al.	Jun 2012	A1
20120158384	Wei	Jun 2012	A1
20120165633	Khair	Jun 2012	A1
20120176190	Goodman et al.	Jul 2012	A1
20120176609	Seppa et al.	Jul 2012	A1
20120217557	Nogami	Aug 2012	A1
20120229206	van Zelm et al.	Sep 2012	A1
20120256687	Davies	Oct 2012	A1
20120259600	Al-Duwaish	Oct 2012	A1
20120263256	Waheed et al.	Oct 2012	A1
20120306573	Mazzucco et al.	Dec 2012	A1
20120328128	Tronchin et al.	Dec 2012	A1
20130005283	van Zelm et al.	Jan 2013	A1
20130009702	Davies	Jan 2013	A1
20130015917	Ishikawa et al.	Jan 2013	A1
20130030239	Weyh et al.	Jan 2013	A1
20130034188	Rashev et al.	Feb 2013	A1
20130040587	Ishikawa et al.	Feb 2013	A1
20130044791	Rimini et al.	Feb 2013	A1
20130044836	Koren et al.	Feb 2013	A1
20130093676	Liberty et al.	Apr 2013	A1
20130110974	Arrasvuori et al.	May 2013	A1
20130113559	Koren et al.	May 2013	A1
20130114762	Azadet et al.	May 2013	A1
20130166259	Weber et al.	Jun 2013	A1
20130170842	Koike-Akino et al.	Jul 2013	A1
20130176153	Rachid et al.	Jul 2013	A1
20130201316	Binder et al.	Aug 2013	A1
20130207723	Chen et al.	Aug 2013	A1
20130222059	Kilambi et al.	Aug 2013	A1
20130243119	Dalipi et al.	Sep 2013	A1
20130243122	Bai	Sep 2013	A1
20130243135	Row et al.	Sep 2013	A1
20130257530	Davies	Oct 2013	A1
20130271212	Bai	Oct 2013	A1
20130272367	Beidas	Oct 2013	A1
20130285742	van Zelm et al.	Oct 2013	A1
20130301487	Khandani	Nov 2013	A1
20130303103	Mikhemar et al.	Nov 2013	A1
20130315291	Kim et al.	Nov 2013	A1
20130321078	Ishikawa et al.	Dec 2013	A1
20130330082	Alonso et al.	Dec 2013	A1
20130336377	Liu et al.	Dec 2013	A1
20140009224	van Zelm et al.	Jan 2014	A1
20140029658	Kim et al.	Jan 2014	A1
20140029660	Bolstad et al.	Jan 2014	A1
20140030995	Kim et al.	Jan 2014	A1
20140031651	Chon	Jan 2014	A1
20140036969	Wyville et al.	Feb 2014	A1
20140044318	Derakhshani et al.	Feb 2014	A1
20140044319	Derakhshani et al.	Feb 2014	A1
20140044320	Derakhshani et al.	Feb 2014	A1
20140044321	Derakhshani et al.	Feb 2014	A1
20140072074	Utsunomiya et al.	Mar 2014	A1
20140077981	Rachid et al.	Mar 2014	A1
20140081157	Joeken	Mar 2014	A1
20140086356	Azadet et al.	Mar 2014	A1
20140086361	Azadet et al.	Mar 2014	A1
20140095129	Liu et al.	Apr 2014	A1
20140107832	Danko	Apr 2014	A1
20140126670	Monsen	May 2014	A1
20140126675	Monsen	May 2014	A1
20140133848	Koike-Akino et al.	May 2014	A1
20140140250	Kim et al.	May 2014	A1
20140161207	Teterwak	Jun 2014	A1
20140167704	Lafontaine et al.	Jun 2014	A1
20140172338	Lafontaine et al.	Jun 2014	A1
20140198959	Derakhshani et al.	Jul 2014	A1
20140213919	Poon et al.	Jul 2014	A1
20140225451	Lafontaine et al.	Aug 2014	A1
20140226035	Nurmenniemi	Aug 2014	A1
20140226828	Klippel et al.	Aug 2014	A1
20140229132	Lafontaine et al.	Aug 2014	A1
20140247906	Pang et al.	Sep 2014	A1
20140266431	Chen	Sep 2014	A1
20140269857	Rimini et al.	Sep 2014	A1
20140269970	Pratt et al.	Sep 2014	A1
20140269989	Santucci et al.	Sep 2014	A1
20140269990	Chen	Sep 2014	A1
20140270405	Derakhshani et al.	Sep 2014	A1
20140278303	Larimore	Sep 2014	A1
20140279778	Lazar et al.	Sep 2014	A1
20140292406	Dechen et al.	Oct 2014	A1
20140292412	Feldman et al.	Oct 2014	A1
20140294119	Sochacki	Oct 2014	A1
20140294252	Derakhshani et al.	Oct 2014	A1
20140313946	Azadet	Oct 2014	A1
20140314176	Azadet	Oct 2014	A1
20140314181	Azadet	Oct 2014	A1
20140314182	Azadet	Oct 2014	A1
20140317163	Azadet et al.	Oct 2014	A1
20140323891	Sterling et al.	Oct 2014	A1
20140333376	Hammi	Nov 2014	A1
20140372091	Larimore	Dec 2014	A1
20150003625	Uhle et al.	Jan 2015	A1
20150005902	van Zelm et al.	Jan 2015	A1
20150016567	Chen	Jan 2015	A1
20150018632	Khair	Jan 2015	A1
20150025328	Khair	Jan 2015	A1
20150031317	Wang	Jan 2015	A1
20150031969	Khair	Jan 2015	A1
20150032788	Velazquez et al.	Jan 2015	A1
20150043678	Hammi	Feb 2015	A1
20150051513	Hunter et al.	Feb 2015	A1
20150061911	Pagnanelli	Mar 2015	A1
20150070089	Eliaz et al.	Mar 2015	A1
20150077180	Matsubara et al.	Mar 2015	A1
20150078484	Xiao et al.	Mar 2015	A1
20150092830	Kim et al.	Apr 2015	A1
20150098710	Agazzi et al.	Apr 2015	A1
20150104196	Bae et al.	Apr 2015	A1
20150131757	Carbone et al.	May 2015	A1
20150146805	Terry	May 2015	A1
20150146806	Terry	May 2015	A1
20150156003	Khandani	Jun 2015	A1
20150156004	Khandani	Jun 2015	A1
20150162881	Hammi	Jun 2015	A1
20150172081	Wloczysiak	Jun 2015	A1
20150180495	Klippel	Jun 2015	A1
20150193565	Kim et al.	Jul 2015	A1
20150193666	Derakhshani et al.	Jul 2015	A1
20150194989	Mkadem et al.	Jul 2015	A1
20150202440	Moehlis et al.	Jul 2015	A1
20150214987	Yu et al.	Jul 2015	A1
20150215937	Khandani	Jul 2015	A1
20150223748	Warrick et al.	Aug 2015	A1
20150230105	Negus et al.	Aug 2015	A1
20150241996	Liberty et al.	Aug 2015	A1
20150249889	Iyer et al.	Sep 2015	A1
20150256216	Ding et al.	Sep 2015	A1
20150270856	Breynaert et al.	Sep 2015	A1
20150270865	Polydoros et al.	Sep 2015	A1
20150280945	Tan et al.	Oct 2015	A1
20150288375	Rachid et al.	Oct 2015	A1
20150295643	Zhao et al.	Oct 2015	A1
20150310739	Beaurepaire et al.	Oct 2015	A1
20150311927	Beidas et al.	Oct 2015	A1
20150311973	Colavolpe et al.	Oct 2015	A1
20150311985	Kim et al.	Oct 2015	A1
20150322647	Danko	Nov 2015	A1
20150326190	Gustavsson	Nov 2015	A1
20150333781	Alon et al.	Nov 2015	A1
20150357975	Avniel et al.	Dec 2015	A1
20150358042	Zhang	Dec 2015	A1
20150358191	Eliaz et al.	Dec 2015	A1
20150381216	Shor et al.	Dec 2015	A1
20150381220	Gal et al.	Dec 2015	A1
20150381821	Kechichian et al.	Dec 2015	A1
20160005419	Chang et al.	Jan 2016	A1
20160022161	Khair	Jan 2016	A1
20160028433	Ding et al.	Jan 2016	A1
20160034421	Magesacher et al.	Feb 2016	A1
20160036472	Chang	Feb 2016	A1
20160036528	Zhao et al.	Feb 2016	A1
20160065311	Winzer et al.	Mar 2016	A1
20160079933	Fehri et al.	Mar 2016	A1
20160087604	Kim et al.	Mar 2016	A1
20160087657	Yu et al.	Mar 2016	A1
20160093029	Micovic et al.	Mar 2016	A1
20160099776	Nakashima et al.	Apr 2016	A1
20160111110	Gautama	Apr 2016	A1
20160117430	Fehri et al.	Apr 2016	A1
20160124903	Agee	May 2016	A1
20160126903	Abdelrahman et al.	May 2016	A1
20160127113	Khandani	May 2016	A1
20160132735	Derakhshani et al.	May 2016	A1
20160134380	Kim et al.	May 2016	A1
20160149665	Henry et al.	May 2016	A1
20160149731	Henry et al.	May 2016	A1
20160156375	Yang et al.	Jun 2016	A1
20160162042	Liberty et al.	Jun 2016	A1
20160173117	Rachid et al.	Jun 2016	A1
20160191020	Velazquez	Jun 2016	A1
20160197642	Henry et al.	Jul 2016	A1
20160218752	Liu	Jul 2016	A1
20160218891	Nammi et al.	Jul 2016	A1
20160225385	Hammarqvist	Aug 2016	A1
20160226681	Henry et al.	Aug 2016	A1
20160241277	Rexberg et al.	Aug 2016	A1
20160248531	Eliaz et al.	Aug 2016	A1
20160259960	Derakhshani et al.	Sep 2016	A1
20160261241	Hammi et al.	Sep 2016	A1
20160269210	Kim et al.	Sep 2016	A1
20160287871	Bardakjian et al.	Oct 2016	A1
20160308619	Ling	Oct 2016	A1
20160309042	Kechichian et al.	Oct 2016	A1
20160316283	Kim et al.	Oct 2016	A1
20160317096	Adams	Nov 2016	A1
20160329927	Xiong et al.	Nov 2016	A1
20160334466	Rivoir	Nov 2016	A1
20160336762	Hunter et al.	Nov 2016	A1
20160352361	Fonseka et al.	Dec 2016	A1
20160352362	Fonseka et al.	Dec 2016	A1
20160352419	Fonseka et al.	Dec 2016	A1
20160352427	Anandakumar et al.	Dec 2016	A1
20160359552	Monsen et al.	Dec 2016	A1
20160373212	Ling et al.	Dec 2016	A1
20160380661	Xue et al.	Dec 2016	A1
20160380700	Shen	Dec 2016	A1
20170012585	Sarbishaei et al.	Jan 2017	A1
20170012709	Duthel	Jan 2017	A1
20170012862	Terry	Jan 2017	A1
20170014032	Khair	Jan 2017	A1
20170018851	Henry et al.	Jan 2017	A1
20170018852	Adriazola et al.	Jan 2017	A1
20170019131	Henry et al.	Jan 2017	A1
20170026095	Ashrafi et al.	Jan 2017	A1
20170032129	Linde et al.	Feb 2017	A1
20170032184	Derakhshani et al.	Feb 2017	A1
20170033465	Henry et al.	Feb 2017	A1
20170033466	Henry et al.	Feb 2017	A1
20170033809	Liu	Feb 2017	A1
20170033953	Henry et al.	Feb 2017	A1
20170033954	Henry et al.	Feb 2017	A1
20170041124	Khandani	Feb 2017	A1
20170043166	Choi et al.	Feb 2017	A1
20170047899	Abdelrahman et al.	Feb 2017	A1
20170061045	Okuyama et al.	Mar 2017	A1
20170063312	Sulimarski et al.	Mar 2017	A1
20170063430	Henry et al.	Mar 2017	A1
20170077944	Pagnanelli	Mar 2017	A1
20170077945	Pagnanelli	Mar 2017	A1
20170078023	Magri et al.	Mar 2017	A1
20170078027	Okabe et al.	Mar 2017	A1
20170078400	Binder et al.	Mar 2017	A1
20170085003	Johnson et al.	Mar 2017	A1
20170085336	Henry et al.	Mar 2017	A1
20170093497	Ling et al.	Mar 2017	A1
20170093693	Barzegar et al.	Mar 2017	A1
20170095195	Hunter et al.	Apr 2017	A1
20170104503	Pratt et al.	Apr 2017	A1
20170104617	Magers	Apr 2017	A1
20170108943	Liberty et al.	Apr 2017	A1
20170110795	Henry et al.	Apr 2017	A1
20170110804	Henry et al.	Apr 2017	A1
20170111805	Barzegar et al.	Apr 2017	A1
20170117854	Ben Smida et al.	Apr 2017	A1
20170134205	Kim et al.	May 2017	A1
20170141807	Chen et al.	May 2017	A1
20170141938	Chen et al.	May 2017	A1
20170163465	Piazza et al.	Jun 2017	A1
20170170999	Zhou et al.	Jun 2017	A1
20170180061	Ishikawa et al.	Jun 2017	A1
20170195053	Duthel	Jul 2017	A1
20170201288	Magers	Jul 2017	A1
20170207934	Iyer Seshadri et al.	Jul 2017	A1
20170214468	Agazzi et al.	Jul 2017	A1
20170214470	Nishihara et al.	Jul 2017	A1
20170222717	Rope	Aug 2017	A1
20170229782	Adriazola et al.	Aug 2017	A1
20170230083	Adriazola et al.	Aug 2017	A1
20170244582	Gal et al.	Aug 2017	A1
20170245054	Sheen et al.	Aug 2017	A1
20170245079	Sheen et al.	Aug 2017	A1
20170245157	Henry et al.	Aug 2017	A1
20170255593	Agee	Sep 2017	A9
20170269481	Borodovsky	Sep 2017	A1
20170271117	Borodovsky	Sep 2017	A1
20170272283	Kingery et al.	Sep 2017	A1
20170288917	Henry et al.	Oct 2017	A1
20170295048	Terry	Oct 2017	A1
20170304625	Buschman	Oct 2017	A1
20170311307	Negus et al.	Oct 2017	A1
20170317781	Henry et al.	Nov 2017	A1
20170317782	Henry et al.	Nov 2017	A1
20170317783	Henry et al.	Nov 2017	A1
20170317858	Henry et al.	Nov 2017	A1
20170318482	Negus et al.	Nov 2017	A1
20170322243	Lafontaine et al.	Nov 2017	A1
20170324421	Tangudu et al.	Nov 2017	A1
20170331899	Binder et al.	Nov 2017	A1
20170338841	Pratt	Nov 2017	A1
20170338842	Pratt	Nov 2017	A1
20170339569	Khandani	Nov 2017	A1
20170346510	Chen et al.	Nov 2017	A1
20170366209	Weissman et al.	Dec 2017	A1
20170366259	Rope	Dec 2017	A1
20170373647	Chang et al.	Dec 2017	A1
20170373759	Rope et al.	Dec 2017	A1
20180013452	Henry et al.	Jan 2018	A9
20180013456	Miyazaki et al.	Jan 2018	A1
20180013495	Ling	Jan 2018	A1
20180026586	Carbone et al.	Jan 2018	A1
20180026673	Kim et al.	Jan 2018	A1
20180034912	Binder et al.	Feb 2018	A1
20180041219	Rachid et al.	Feb 2018	A1
20180048497	Henry et al.	Feb 2018	A1
20180054232	Henry et al.	Feb 2018	A1
20180054233	Henry et al.	Feb 2018	A1
20180054234	Stuckman et al.	Feb 2018	A1
20180054268	Abdoli et al.	Feb 2018	A1
20180062674	Boghrat et al.	Mar 2018	A1
20180062886	Henry et al.	Mar 2018	A1
20180069594	Henry et al.	Mar 2018	A1
20180069731	Henry et al.	Mar 2018	A1
20180070394	Khandani	Mar 2018	A1
20180076947	Kazakevich et al.	Mar 2018	A1
20180076979	Lomayev et al.	Mar 2018	A1
20180076982	Henry et al.	Mar 2018	A1
20180076988	Willis et al.	Mar 2018	A1
20180091195	Carvalho et al.	Mar 2018	A1
20180102850	Agazzi et al.	Apr 2018	A1
20180115040	Bennett et al.	Apr 2018	A1
20180115058	Henry et al.	Apr 2018	A1
20180123256	Henry et al.	May 2018	A1
20180123257	Henry et al.	May 2018	A1
20180123749	Azizi et al.	May 2018	A1
20180123836	Henry et al.	May 2018	A1
20180123856	Qu et al.	May 2018	A1
20180123897	Lee et al.	May 2018	A1
20180124181	Binder et al.	May 2018	A1
20180131406	Adriazola et al.	May 2018	A1
20180131502	Askar et al.	May 2018	A1
20180131541	Henry et al.	May 2018	A1
20180145411	Henry et al.	May 2018	A1
20180145412	Henry et al.	May 2018	A1
20180145414	Henry et al.	May 2018	A1
20180145415	Henry et al.	May 2018	A1
20180151957	Bennett et al.	May 2018	A1
20180152262	Henry et al.	May 2018	A1
20180152330	Chritz et al.	May 2018	A1
20180152925	Nammi et al.	May 2018	A1
20180159195	Henry et al.	Jun 2018	A1
20180159196	Henry et al.	Jun 2018	A1
20180159197	Henry et al.	Jun 2018	A1
20180159228	Britz et al.	Jun 2018	A1
20180159229	Britz	Jun 2018	A1
20180159230	Henry et al.	Jun 2018	A1
20180159232	Henry et al.	Jun 2018	A1
20180159240	Henry et al.	Jun 2018	A1
20180159243	Britz et al.	Jun 2018	A1
20180159615	Kim	Jun 2018	A1
20180166761	Henry et al.	Jun 2018	A1
20180166784	Johnson et al.	Jun 2018	A1
20180166785	Henry et al.	Jun 2018	A1
20180166787	Johnson et al.	Jun 2018	A1
20180167042	Nagasaku	Jun 2018	A1
20180167092	Hausmair et al.	Jun 2018	A1
20180167093	Miyazaki et al.	Jun 2018	A1
20180167105	Vannucci et al.	Jun 2018	A1
20180167148	Vannucci et al.	Jun 2018	A1
20180175892	Henry et al.	Jun 2018	A1
20180175978	Beidas et al.	Jun 2018	A1
20180180420	Korenberg	Jun 2018	A1
20180191448	Kaneda et al.	Jul 2018	A1
20180198668	Kim et al.	Jul 2018	A1
20180205399	Baringer et al.	Jul 2018	A1
20180205481	Shlomo et al.	Jul 2018	A1
20180219566	Weissman et al.	Aug 2018	A1
20180227158	Luo et al.	Aug 2018	A1
20180248592	Ashrafi	Aug 2018	A1
20180254754	Carvalho et al.	Sep 2018	A1
20180254769	Alic et al.	Sep 2018	A1
20180254924	Berardinelli et al.	Sep 2018	A1
20180262217	Hassan	Sep 2018	A1
20180262243	Ashrafi et al.	Sep 2018	A1
20180262370	Al-Mufti et al.	Sep 2018	A1
20180269988	Yu et al.	Sep 2018	A1
20180278693	Binder et al.	Sep 2018	A1
20180278694	Binder et al.	Sep 2018	A1
20180279197	Kim et al.	Sep 2018	A1
20180294879	Anandakumar et al.	Oct 2018	A1
20180294884	Rope et al.	Oct 2018	A1
20180294897	Vannucci et al.	Oct 2018	A1
20180301812	Henry et al.	Oct 2018	A1
20180302111	Chen et al.	Oct 2018	A1
20180302145	Kim	Oct 2018	A1
20180309206	Henry et al.	Oct 2018	A1
20180309465	Pratt	Oct 2018	A1
20180316320	Soubercaze-Pun et al.	Nov 2018	A1
20180323826	Vannucci et al.	Nov 2018	A1
20180324005	Kim et al.	Nov 2018	A1
20180324006	Henry et al.	Nov 2018	A1
20180324021	Chritz et al.	Nov 2018	A1
20180324601	Barzegar et al.	Nov 2018	A1
20180331413	Adriazola et al.	Nov 2018	A1
20180331720	Adriazola et al.	Nov 2018	A1
20180331721	Adriazola et al.	Nov 2018	A1
20180331814	Khandani	Nov 2018	A1
20180331871	Martinez	Nov 2018	A1
20180333580	Choi et al.	Nov 2018	A1
20180343304	Binder et al.	Nov 2018	A1
20180351687	Henry et al.	Dec 2018	A1
20180358678	Henry et al.	Dec 2018	A1
20180359126	Wang	Dec 2018	A1
20180367219	Dar et al.	Dec 2018	A1
20180375940	Binder et al.	Dec 2018	A1
20190007075	Kim et al.	Jan 2019	A1
20190013577	Henry et al.	Jan 2019	A1
20190013837	Henry et al.	Jan 2019	A1
20190013838	Henry et al.	Jan 2019	A1
20190013867	Ling et al.	Jan 2019	A1
20190013874	Ling	Jan 2019	A1
20190013991	Duyck et al.	Jan 2019	A1
20190020415	Agazzi et al.	Jan 2019	A1
20190020530	Au et al.	Jan 2019	A1
20190028131	Wang	Jan 2019	A1
20190030334	Lerman et al.	Jan 2019	A1
20190036222	Henry et al.	Jan 2019	A1
20190036622	Lagoy et al.	Jan 2019	A1
20190042536	Agee	Feb 2019	A1
20190052505	Baldemair et al.	Feb 2019	A1
20190074563	Henry et al.	Mar 2019	A1
20190074564	Henry et al.	Mar 2019	A1
20190074565	Henry et al.	Mar 2019	A1
20190074568	Henry et al.	Mar 2019	A1
20190074580	Henry et al.	Mar 2019	A1
20190074584	Henry et al.	Mar 2019	A1
20190074597	Vannucci et al.	Mar 2019	A1
20190074598	Henry et al.	Mar 2019	A1
20190074864	Henry et al.	Mar 2019	A1
20190074865	Henry et al.	Mar 2019	A1
20190074878	Henry et al.	Mar 2019	A1
20190109568	Ben Smida et al.	Apr 2019	A1
20190115877	Ben Smida et al.	Apr 2019	A1
20190150774	Brinkmann et al.	May 2019	A1
20190199365	Rachid et al.	Jun 2019	A1
20190204369	Lafontaine et al.	Jul 2019	A1
20190207589	Alic et al.	Jul 2019	A1
20190215073	Schmogrow et al.	Jul 2019	A1
20190222179	Doi	Jul 2019	A1
20190222326	Dunworth et al.	Jul 2019	A1
20190260401	Megretski et al.	Aug 2019	A1
20190260402	Chuang et al.	Aug 2019	A1
20190268026	Liu	Aug 2019	A1
20190280730	Zhang et al.	Sep 2019	A1
20190280778	Zhu et al.	Sep 2019	A1
20190312552	Chan et al.	Oct 2019	A1
20190312571	Hovakimyan et al.	Oct 2019	A1
20190312648	Cavaliere et al.	Oct 2019	A1
20190326942	Spring et al.	Oct 2019	A1
20190348956	Megretski et al.	Nov 2019	A1
20190386621	Alon et al.	Dec 2019	A1
20190386750	Wang et al.	Dec 2019	A1
20200000355	Khair	Jan 2020	A1
20200008686	Khair	Jan 2020	A1
20200012754	Larimore	Jan 2020	A1
20200028476	Kim et al.	Jan 2020	A1
20200064406	Ritzberger et al.	Feb 2020	A1
20200067543	Kim et al.	Feb 2020	A1
20200067600	Agazzi et al.	Feb 2020	A1
20200076379	Tanio et al.	Mar 2020	A1
20200091608	Alpman et al.	Mar 2020	A1
20200110474	Liberty et al.	Apr 2020	A1
20200119756	Boghrat et al.	Apr 2020	A1
20200145033	Rafique	May 2020	A1
20200145112	Wang et al.	May 2020	A1
20200162035	Gorbachov	May 2020	A1
20200169227	Megretski et al.	May 2020	A1
20200169334	Li et al.	May 2020	A1
20200169937	Kim et al.	May 2020	A1
20200174514	Amiralizadeh Asl et al.	Jun 2020	A1
20200186103	Weber et al.	Jun 2020	A1
20200195296	Sarkas et al.	Jun 2020	A1
20200225308	Dosenbach et al.	Jul 2020	A1
20200244232	Cope et al.	Jul 2020	A1
20200252032	Faig et al.	Aug 2020	A1
20200259465	Wu et al.	Aug 2020	A1
20200266768	Chen et al.	Aug 2020	A1
20200274560	Pratt	Aug 2020	A1
20200279546	Chase	Sep 2020	A1
20200295975	Li et al.	Sep 2020	A1
20200313763	Wang et al.	Oct 2020	A1
20200321920	Chiron	Oct 2020	A1
20200333295	Schiffres et al.	Oct 2020	A1
20200351072	Khandani	Nov 2020	A1
20200366253	Megretski et al.	Nov 2020	A1
20200382147	Menkhoff et al.	Dec 2020	A1
20200389207	Taniguchi et al.	Dec 2020	A1
20200395662	Tervo et al.	Dec 2020	A1
20210006207	Carvalho et al.	Jan 2021	A1
20210013843	Cohen et al.	Jan 2021	A1
20210021238	Chan et al.	Jan 2021	A1
20210044461	Groen et al.	Feb 2021	A1
20210075464	Luo	Mar 2021	A1
20210075649	Mahmood et al.	Mar 2021	A1
20210097128	Murase	Apr 2021	A1

Non-Patent Literature Citations (341)

Entry
U.S. Appl. No. 10/003,364, filed Jun. 19, 2018, Willis III et al.
U.S. Appl. No. 10/008,218, filed Jun. 26, 2018, Wu et al.
U.S. Appl. No. 10/009,050, filed Jun. 26, 2018, Chen et al.
U.S. Appl. No. 10/009,109, filed Jun. 26, 2018, Rope et al.
U.S. Appl. No. 10/009,259, filed Jun. 26, 2018, Calmon et al.
U.S. Appl. No. 10/013,515, filed Jul. 3, 2018, Williams.
U.S. Appl. No. 10/015,593, filed Jul. 3, 2018, Iyer et al.
U.S. Appl. No. 10/027,397, filed Jul. 17, 2018, Kim.
U.S. Appl. No. 10/027,427, filed Jul. 17, 2018, Vannucci et al.
U.S. Appl. No. 10/027,523, filed Jul. 17, 2018, Chritz et al.
U.S. Appl. No. 10/033,107, filed Jul. 24, 2018, Henry et al.
U.S. Appl. No. 10/033,108, filed Jul. 24, 2018, Henry et al.
U.S. Appl. No. 10/033,413, filed Jul. 24, 2018, Pratt.
U.S. Appl. No. 10/033,568, filed Jul. 24, 2018, Piazza et al.
U.S. Appl. No. 10/050,636, filed Aug. 14, 2018, Tangudu et al.
U.S. Appl. No. 10/050,710, filed Aug. 14, 2018, Anandakumar et al.
U.S. Appl. No. 10/050,714, filed Aug. 14, 2018, Nishihara et al.
U.S. Appl. No. 10/050,815, filed Aug. 14, 2018, Henry et al.
U.S. Appl. No. 10/051,483, filed Aug. 14, 2018, Barzegar et al.
U.S. Appl. No. 10/051,488, filed Aug. 14, 2018, Vannucci et al.
U.S. Appl. No. 10/062,970, filed Aug. 28, 2018, Vannucci et al.
U.S. Appl. No. 10/063,265, filed Aug. 28, 2018, Pratt et al.
U.S. Appl. No. 10/063,354, filed Aug. 28, 2018, Hadani et al.
U.S. Appl. No. 10/063,364, filed Aug. 28, 2018, Khandani.
U.S. Appl. No. 10/069,467, filed Sep. 4, 2018, Carvalho et al.
U.S. Appl. No. 10/069,535, filed Sep. 4, 2018, Vannucci et al.
U.S. Appl. No. 10/075,201, filed Sep. 11, 2018, Gazneli et al.
U.S. Appl. No. 10/079,652, filed Sep. 18, 2018, Henry et al.
U.S. Appl. No. 10/084,562, filed Sep. 25, 2018, Abdoli et al.
U.S. Appl. No. 10/090,594, filed Oct. 2, 2018, Henry et al.
U.S. Appl. No. 10/095,927, filed Oct. 9, 2018, Derakhshani et al.
U.S. Appl. No. 10/096,883, filed Oct. 9, 2018, Henry et al.
U.S. Appl. No. 10/097,273, filed Oct. 9, 2018, Agazzi et al.
U.S. Appl. No. 10/097,939, filed Oct. 9, 2018, Sheen et al.
U.S. Appl. No. 10/101,370, filed Oct. 16, 2018, Lafontaine et al.
U.S. Appl. No. 10/103,777, filed Oct. 16, 2018, Henry et al.
U.S. Appl. No. 10/108,858, filed Oct. 23, 2018, Derakhshani et al.
U.S. Appl. No. 10/110,315, filed Oct. 23, 2018, Ling.
U.S. Appl. No. 10/116,390, filed Oct. 30, 2018, Ling et al.
U.S. Appl. No. 10/123,217, filed Nov. 6, 2018, Barzegar et al.
U.S. Appl. No. 10/128,955, filed Nov. 13, 2018, Rope et al.
U.S. Appl. No. 10/129,057, filed Nov. 13, 2018, Willis III et al.
U.S. Appl. No. 10/135,145, filed Nov. 20, 2018, Henry et al.
U.S. Appl. No. 10/141,944, filed Nov. 27, 2018, Rachid et al.
U.S. Appl. No. 10/142,754, filed Nov. 27, 2018, Sheen et al.
U.S. Appl. No. 10/147,431, filed Dec. 4, 2018, Dick et al.
U.S. Appl. No. 10/148,016, filed Dec. 4, 2018, Johnson et al.
U.S. Appl. No. 10/148,360, filed Dec. 4, 2018, Ashrafi.
U.S. Appl. No. 10/148,417, filed Dec. 4, 2018, Ling et al.
U.S. Appl. No. 10/153,793, filed Dec. 11, 2018, Hausmair et al.
U.S. Appl. No. 10/168,501, filed Jan. 1, 2019, Ashrafi.
U.S. Appl. No. 10/170,840, filed Jan. 1, 2019, Henry et al.
U.S. Appl. No. 10/171,158, filed Jan. 1, 2019, Barzegar et al.
U.S. Appl. No. 10/181,825, filed Jan. 15, 2019, Ben Smida et al.
U.S. Appl. No. 10/191,376, filed Jan. 29, 2019, Borodovsky.
U.S. Appl. No. 10/198,582, filed Feb. 5, 2019, Linde et al.
U.S. Appl. No. 10/200,106, filed Feb. 5, 2019, Barzegar et al.
U.S. Appl. No. 10/205,212, filed Feb. 12, 2019, Henry et al.
U.S. Appl. No. 10/205,231, filed Feb. 12, 2019, Henry et al.
U.S. Appl. No. 10/205,482, filed Feb. 12, 2019, Barzegar et al.
U.S. Appl. No. 10/205,655, filed Feb. 12, 2019, Barzegar et al.
U.S. Appl. No. 10/211,855, filed Feb. 19, 2019, Baringer et al.
U.S. Appl. No. 10/212,014, filed Feb. 19, 2019, Qu et al.
U.S. Appl. No. 10/218,405, filed Feb. 26, 2019, Magers.
U.S. Appl. No. 10/224,634, filed Mar. 5, 2019, Henry et al.
U.S. Appl. No. 10/224,970, filed Mar. 5, 2019, Pratt.
U.S. Appl. No. 10/230,353, filed Mar. 12, 2019, Alic et al.
U.S. Appl. No. 10/230,550, filed Mar. 12, 2019, Al-Mufti et al.
U.S. Appl. No. 10/237,114, filed Mar. 19, 2019, Duyck et al.
U.S. Appl. No. 10/243,525, filed Mar. 26, 2019, Nagasaku.
U.S. Appl. No. 10/243,596, filed Mar. 26, 2019, Kerhuel et al.
U.S. Appl. No. 10/268,169, filed Apr. 23, 2019, van Zelm et al.
U.S. Appl. No. 10/270,478, filed Apr. 23, 2019, Liu.
U.S. Appl. No. 10/291,384, filed May 14, 2019, Askar et al.
U.S. Appl. No. 10/305,432, filed May 28, 2019, Trayling et al.
U.S. Appl. No. 10/305,435, filed May 28, 2019, Murugesu et al.
U.S. Appl. No. 10/311,243, filed Jun. 4, 2019, Calmon et al.
U.S. Appl. No. 10/320,340, filed Jun. 11, 2019, Pratt et al.
U.S. Appl. No. 10/333,474, filed Jun. 25, 2019, Alon et al.
U.S. Appl. No. 10/333,561, filed Jun. 25, 2019, Liu.
U.S. Appl. No. 10/333,764, filed Jun. 25, 2019, Arditti Ilitzky.
U.S. Appl. No. 10/334,637, filed Jun. 25, 2019, Khandani.
U.S. Appl. No. 10/348,341, filed Jul. 9, 2019, Weissman et al.
U.S. Appl. No. 10/348,345, filed Jul. 9, 2019, Guyton et al.
U.S. Appl. No. 10/374,781, filed Aug. 6, 2019, Khandani.
U.S. Appl. No. 10/384,064, filed Aug. 20, 2019, Choi et al.
U.S. Appl. No. 10/389,449, filed Aug. 20, 2019, Ling et al.
U.S. Appl. No. 10/396,723, filed Aug. 27, 2019, Haas et al.
U.S. Appl. No. 10/404,296, filed Sep. 3, 2019, Kim et al.
U.S. Appl. No. 10/404,376, filed Sep. 3, 2019, Schmogrow et al.
U.S. Appl. No. 10/417,353, filed Sep. 17, 2019, Larimore.
U.S. Appl. No. 10/419,046, filed Sep. 17, 2019, Chen et al.
U.S. Appl. No. 10/419,126, filed Sep. 17, 2019, Ling.
U.S. Appl. No. 10/447,211, filed Oct. 15, 2019, Rollins et al.
U.S. Appl. No. 10/447,244, filed Oct. 15, 2019, Zhou et al.
U.S. Appl. No. 10/452,621, filed Oct. 22, 2019, Medard et al.
U.S. Appl. No. 10/463,276, filed Nov. 5, 2019, Hunter et al.
U.S. Appl. No. 10/474,775, filed Nov. 12, 2019, Okuyama et al.
U.S. Appl. No. 10/491,170, filed Nov. 26, 2019, Ben Smida et al.
U.S. Appl. No. 10/491,171, filed Nov. 26, 2019, Ben Smida et al.
U.S. Appl. No. 10/498,372, filed Dec. 3, 2019, Pratt.
U.S. Appl. No. 10/505,638, filed Dec. 10, 2019, Agazzi et al.
U.S. Appl. No. 10/511,337, filed Dec. 17, 2019, Boghrat et al.
U.S. Appl. No. 10/514,776, filed Dec. 24, 2019, Liberty et al.
U.S. Appl. No. 10/516,434, filed Dec. 24, 2019, Chang et al.
U.S. Appl. No. 10/523,159, filed Dec. 31, 2019, Megretski et al.
U.S. Appl. No. 10/530,574, filed Jan. 7, 2020, Shi et al.
U.S. Appl. No. 10/534,325, filed Jan. 14, 2020, Dash et al.
U.S. Appl. No. 10/536,221, filed Jan. 14, 2020, Wang et al.
U.S. Appl. No. 10/540,984, filed Jan. 21, 2020, Malik et al.
U.S. Appl. No. 10/545,482, filed Jan. 28, 2020, Dash et al.
U.S. Appl. No. 10/554,183, filed Feb. 4, 2020, Doi.
U.S. Appl. No. 10/560,140, filed Feb. 11, 2020, Wu et al.
U.S. Appl. No. 10/581,470, filed Mar. 3, 2020, Megretski et al.
U.S. Appl. No. 10/594,406, filed Mar. 17, 2020, Zhu et al.
U.S. Appl. No. 10/608,751, filed Mar. 31, 2020, Yu et al.
U.S. Appl. No. 10/622,951, filed Apr. 14, 2020, Chen et al.
U.S. Appl. No. 10/623,049, filed Apr. 14, 2020, Zhang et al.
U.S. Appl. No. 10/623,118, filed Apr. 14, 2020, Lagoy et al.
U.S. Appl. No. 10/630,323, filed Apr. 21, 2020, Spring et al.
U.S. Appl. No. 10/630,391, filed Apr. 21, 2020, LaGasse et al.
U.S. Appl. No. 10/644,657, filed May 5, 2020, Megretski et al.
U.S. Appl. No. 10/666,307, filed May 26, 2020, Wang.
U.S. Appl. No. 10/735,032, filed Aug. 4, 2020, Liu.
U.S. Appl. No. 10/741,188, filed Aug. 11, 2020, Dick et al.
U.S. Appl. No. 10/742,388, filed Aug. 11, 2020, Khandani.
U.S. Appl. No. 10/749,480, filed Aug. 18, 2020, Tanio et al.
U.S. Appl. No. 10/756,682, filed Aug. 25, 2020, Faig et al.
U.S. Appl. No. 10/756,774, filed Aug. 25, 2020, Sarkas et al.
U.S. Appl. No. 10/756,822, filed Aug. 25, 2020, Marsella et al.
U.S. Appl. No. 10/763,904, filed Sep. 1, 2020, Megretski et al.
U.S. Appl. No. 10/770,080, filed Sep. 8, 2020, Dick et al.
U.S. Appl. No. 10/775,437, filed Sep. 15, 2020, Rivoir.
U.S. Appl. No. 10/812,166, filed Oct. 20, 2020, Kim et al.
U.S. Appl. No. 10/826,444, filed Nov. 3, 2020, Soubercaze-Pun et al.
U.S. Appl. No. 10/826,615, filed Nov. 3, 2020, Cavaliere et al.
U.S. Appl. No. 10/833,634, filed Nov. 10, 2020, Chan et al.
U.S. Appl. No. 10/841,013, filed Nov. 17, 2020, Agazzi et al.
U.S. Appl. No. 10/862,517, filed Dec. 8, 2020, Kim et al.
U.S. Appl. No. 10/873,301, filed Dec. 22, 2020, Chiron.
U.S. Appl. No. 10/887,022, filed Jan. 5, 2021, Dar et al.
U.S. Appl. No. 10/892,786, filed Jan. 12, 2021, Pratt et al.
U.S. Appl. No. 10/897,276, filed Jan. 19, 2021, Megretski et al.
U.S. Appl. No. 10/902,086, filed Jan. 26, 2021, Agee.
U.S. Appl. No. 10/905,342, filed Feb. 2, 2021, Sterling et al.
U.S. Appl. No. 10/911,029, filed Feb. 2, 2021, Velazquez et al.
U.S. Appl. No. 10/931,238, filed Feb. 23, 2021, Megretski et al.
U.S. Appl. No. 10/931,318, filed Feb. 23, 2021, Mahmood et al.
U.S. Appl. No. 10/931,320, filed Feb. 23, 2021, Megretski et al.
U.S. Appl. No. 10/931,366, filed Feb. 23, 2021, Wang et al.
U.S. Appl. No. 10/965,380, filed Mar. 30, 2021, Kim et al.
U.S. Appl. No. 10/972,139, filed Apr. 6, 2021, Luo.
U.S. Appl. No. 10/979,090, filed Apr. 13, 2021, Rafique.
U.S. Appl. No. 10/979,097, filed Apr. 13, 2021, Luo.
U.S. Appl. No. 10/985,951, filed Apr. 20, 2021, Li et al.
Abdelaziz, M., L. Anttila, and M. Valkama, “Reduced-complexity digital predistortion for massive mimo,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 6478-6482.
Abrudan, Traian. Volterra Series and Nonlinear Adaptive Filters. Technical report, Helsinki University of Technology, 2003.
Aghvami, A. H. and Robertson, I. D. [Apr. 1993] Power limitation and high-power amplifier non linearities in on-board satellite communications systems. Electron. and Comm. Engin. J.
Akaiwa, Y. Introduction to Digital Mobile Communication. New York: Wiley, 1997.
Alexander, N. A., A. A. Chanerley, A. J. Crewe, and S. Bhattacharya. “Obtaining spectrum matching time series using a Reweighted Volterra Series Algorithm (RVSA).” Bulletin of the Seismological Society of America 104, No. 4 (2014): 1663-1673.
Altera Corporation. Digital Predistortion Reference Design. Application Note 314, 2003.
Ahluwalia, Sehej Swaran. “An FPGA Implementation Of Adaptive Linearization Of Power Amplifiers.” PhD diss., Georgia Institute of Technology, 2019.
Akaiwa, Yoshihiko. Introduction to digital mobile communication. John Wiley & Sons, 2015.
Annabestani, Mohsen, and Nadia Naghavi. “Practical realization of discrete-time Volterra series for high-order nonlinearities.” Nonlinear Dynamics 98, No. 3 (2019): 2309-2325.
Arthanayake, T. and Wood, H. B. [Apr. 8, 1971] Linear amplification using envelope feedback. Elec. Lett.
Aysal, Tuncer C., and Kenneth E. Barner, “Myriad-Type Polynomial Filtering”, IEEE Transactions on Signal Processing, vol. 55, No. 2, Feb. 2007.
Barner, Kenneth E., and Tuncer Can Aysal, “Polynomial Weighted Median Filtering”, IEEE Transactions on Signal Processing, vol. 54, No. 2, Feb. 2006.
Barner, Kenneth E,.Gonzalo R. Arce, Giovanni L. Sicuranza, and Ilya Shmulevich, “Nonlinear Signal and Image Processing—Part I”, EURASIP J. Applied Signal Processing (2002).
Barner, Kenneth E,.Gonzalo R. Arce, Giovanni L. Sicuranza, and Ilya Shmulevich, “Nonlinear Signal and Image Processing—Part II”, EURASIP J. Applied Signal Processing (2002).
Bedrosian, Edward, and Stephen O. Rice. “The output properties of Volterra systems (nonlinear systems with memory) driven by harmonic and Gaussian inputs.” Proceedings of the IEEE 59, No. 12 (1971): 1688-1707.
Bennet, T. J. and Clements, R. F. [May 1974] Feedforward—An alternative approach to amplifier linearisation. Radio and Electron. Engin.
Bhargava, V. K. et al. [1981] Digital Communications by Satellite, John Wiley and Sons.
Biglieri, Ezio, Sergio Barberis, and Maurizio Catena, “Analysis and Compensation of Nonlinearities in Digital Transmission Systems”, IEEE Journal on selected areas in Communications, vol. 6, No. 1, Jan. 1988.
Li, Bin, Chenglin Zhao, Mengwei Sun, Haijun Zhang, Zheng Zhou, and Arumugam Nallanathan. “A Bayesian approach for nonlinear equalization and signal detection in millimeter-wave communications.” IEEE Transactions on Wireless Communications 14, No. 7 (2015): 3794-3809.
Birpoutsoukis, Georgios, Anna Marconato, John Lataire, and Johan Schoukens. “Regularized nonparametric Volterra kernel estimation.” Automatica 82 (2017): 324-327.
Black, H. S. [Dec. 1937] Wave translating system. U.S. Pat. No. 2,102,671.
Black, H. S. [Oct. 1928] Translating system. U.S. Pat. No. 1,686,792.
Bohm, D. The Special Theory of Relativity, Benjamin, 1965.
Bond F. E. and Meyer, H. F. [Apr. 1970] Intermodulation effects in limiter amplifier repeaters. IEEE Trans. Comm., vol. COM-18, p. 127-135.
Bouzrara, Kais, Abdelkader Mbarek, and Tarek Garna. “Non-linear predictive controller for uncertain process modelled by GOBF-Volterra models.” International Journal of Modelling, Identification and Control 19, No. 4 (2013): 307-322.
Boyd, Stephen, Leon O. Chua, and Charles A. Desoer. “Analytical foundations of Volterra series.” IMA Journal of Mathematical Control and Information 1, No. 3 (1984): 243-282.
Budura, Georgeta, and Corina Botoca, “Efficient Implementation of the Third Order RLS Adaptive Volterra Filter”, FACTA Universitatis (NIS) Ser.: Elec. Energ. vol. 19, No. 1, Apr. 2006.
Budura, Georgeta, and Corina Botoca. “Nonlinearities identification using the LMS Volterra filter.” Communications Department Faculty of Electronics and Telecommunications Timisoara, Bd. V. Parvan 2 (2005).
Campello, Ricardo JGB, Gérard Favier, and Wagner C. Do Amaral. “Optimal expansions of discrete-time Volterra models using Laguerre functions.” Automatica 40, No. 5 (2004): 815-822.
Campello, Ricardo JGB, Wagner C. Do Amaral, and Gérard Favier. “A note on the optimal expansion of Volterra models using Laguerre functions.” Automatica 42, No. 4 (2006): 689-693.
Carassale, Luigi, and Ahsan Kareem. “Modeling nonlinear systems by Volterra series.” Journal of engineering mechanics 136, No. 6 (2010): 801-818.
Censor, D., & Melamed, T, 2002, Volterra differential constitutive operators and locality considerations in electromagnetic theory, PIER—Progress in Electromagnetic Research, 36: 121-137.
Censor, D., 2000, A quest for systematic constitutive formulations for general field and wave systems based on the Volterra differential operators, PIER—Progress In Electromagnetics Research, (25): 261-284.
Censor, D., 2001, Constitutive relations in inhomogeneous systems and the particle-field conundrum, PIER—Progress In Electromagnetics Research, (30): 305-335.
Censor, D., “Volterra Series and Operators” (2003).
Cheaito, A., M. Crussière, J .- F. Hélard, and Y. Louët, “Quantifying the memory effects of power amplifiers: Evm closed-form derivations of multicarrier signals.” IEEE Wireless Commun. Letters, vol. 6, No. 1, pp. 34-37, 2017.
Chen, S., B. Mulgrew, and P. M. Grant, “A clustering technique for digital communications channel equalization using radial basis function networks,” IEEE Transactions on neural networks, vol. 4, No. 4, pp. 570-590, 1993.
Chen, Hai-Wen. “Modeling and identification of parallel and feedback nonlinear systems.” In Proceedings of 1994 33rd IEEE Conference on Decision and Control, vol. 3, pp. 2267-2272. IEEE, 1994.
Da Rosa, Alex, Ricardo JGB Campello, and Wagner C. Amaral. “Choice of free parameters in expansions of discrete-time Volterra models using Kautz functions.” Automatica 43, No. 6 (2007): 1084-1091.
Deng, Yongjun, and Zhixing Yang. “Comments on” Complex-bilinear recurrent neural network for equalization of a digital Satellite Channel“.” IEEE Transactions on Neural Networks 17, No. 1 (2006): 268.
Dimitrov, S., “Non-linear distortion cancellation and symbol-based equalization in satellite forward links,” IEEE Trans Wireless Commun, vol. 16, No. 7, pp. 4489-4502, 2017.
Ding, L., G. T. Zhou, D. R. Morgan, Z. Ma, J. S. Kenney, J. Kim, and C. R. Giardina, “A robust digital baseband predistorter constructed using memory polynomials,” IEEE Transactions on communications, vol. 52, No. 1, pp. 159-165, 2004.
Doyle III, Francis J., Babatunde A. Ogunnaike, and Ronald K. Pearson. “Nonlinear model-based control using second-order Volterra models.” Automatica 31, No. 5 (1995): 697-714.
ETSI [Aug. 1994] Standard ETR 132. Radio broadcasting systems; Code of practice for site engineering VHF FM sound broadcasting transmitters. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.
ETSI [Jan. 1995] European Standard ETS 300 384. Radio broadcasting systems; Very high frequency (VHF), frequency modulated, sound broadcasting transmitters. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.
ETSI [Jun. 1998] Standard ETR 053 Ed 3—Radio site engineering for equipment and systems in the mobile service. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.
ETSI [Mar. 1997] European Standard ETS 300 113. Radio equipment and systems (RES); Land mobile service; Technical characteristics and test conditions for radio equipment intended for the transmission of data (and speech) and having an antenna connector. European Telecommunications Standards Institute, Sophia Antipolis, F-06291, Valbonne Cedex, France.
Eun, Changsoo, and Edward J. Powers. “A new Volterra predistorter based on the indirect learning architecture.” IEEE transactions on signal processing 45, No. 1 (1997): 223-227.
Evans, Ceri, David Rees, Lee Jones, and Michael Weiss. “Periodic signals for measuring nonlinear Volterra kernels.” IEEE transactions on instrumentation and measurement 45, No. 2 (1996): 362-371.
Fang, Yang-Wang, Li-Cheng Jiao, Xian-Da Zhang and Jin Pan, “On the Convergence of Volterra Filter Equalizers Using a Pth-Order Inverse Approach”, IEEE Transactions on Signal Processing, vol. 49, No. 8, Aug. 2001.
Fang, Xi, Yixin Fu, Xin Sui, Lei Zhang, Xianwei Gao, Ding Ding, and Lifang Liu. “Volterra-based fiber nonlinearity impairment modeling for OFDM/OQAM systems.” In 2019 18th International Conference on Optical Communications and Networks (ICOCN), pp. 1-3. IEEE, 2019.
Fang, Yang-Wang, Li-Cheng Jiao, Xian-Da Zhang, and Jin Pan. “On the convergence of Volterra filter equalizers using a pth-order inverse approach.” IEEE transactions on signal processing 49, No. 8 (2001): 1734-1744.
Frank, Walter A. “An efficient approximation to the quadratic Volterra filter and its application in real-time loudspeaker linearization.” Signal Processing 45, No. 1 (1995): 97-113.
Frank, Walter A. “Sampling requirements for Volterra system identification.” IEEE Signal Processing Letters 3, No. 9 (1996): 266-268.
Franz, Matthias O., and Bernhard Schölkopf. “A unifying view of Wiener and Volterra theory and polynomial kernel regression.” Neural computation 18, No. 12 (2006): 3097-3118.
Franz, Matthias O., Volterra and Weiner Series, Scholarpedia, 6(10):11307 doi: 10.4249/scholarpedia.11307.
Gardner, William A., and Teri L. Archer. “Exploitation of cyclostationarity for identifying the Volterra kernels of nonlinear systems.” IEEE transactions on Information Theory 39, No. 2 (1993): 535-542.
Gardner, William A., and Teri L. Archer. “Simplified methods for identifying the Volterra kernels of nonlinear systems.” In [1991] Proceedings of the 34th Midwest Symposium on Circuits and Systems, pp. 98-101. IEEE, 1992.
Genecili, Hasmet, and Michael Nikolaou. “Design of robust constrained model-predictive controllers with volterra series.” AIChE Journal 41, No. 9 (1995): 2098-2107.
Ghannouchi, Fadhel M., Oualid Hammi, and Mohamed Helaoui. Behavioral modeling and predistortion of wideband wireless transmitters. John Wiley & Sons, 2015.
Gilbert, Elmer. “Functional expansions for the response of nonlinear differential systems.” IEEE transactions on Automatic Control 22, No. 6 (1977): 909-921.
Gray, L. F. [1980] Application of broadband linearisers to satellite transponders. IEEE Conf. Proc. ICC'80.
Guan, L., and A. Zhu, “Simplified dynamic deviation reduction-based volterra model for doherty power amplifiers,” in Integrated Nonlinear Microwave and Millimetre-Wave Circuits (INMMIC), 2011 Workshop on. IEEE, 2011, pp. 1-4.
Guan, Lei, and Anding Zhu. “Low-cost FPGA implementation of Volterra series-based digital predistorter for RF power amplifiers.” IEEE Transactions on Microwave Theory and Techniques 58, No. 4 (2010): 866-872.
Guérin, Alexandre, Gérard Faucon, and Régine Le Bouquin-Jeannès, “Nonlinear Acoustic Echo Cancellation Based on Volterra Filters”, IEEE Transactions on Speech and Audio Processing, vol. 11, No. 6, Nov. 2003.
Guiomar, Fernando P., Jacklyn D. Reis, António L. Teixeira, and Armando N. Pinto. “Mitigation of intra-channel nonlinearities using a frequency-domain Volterra series equalizer.” Optics express 20, No. 2 (2012): 1360-1369.
Hanson, Joshua. “Learning Volterra Series via RKHS Methods.” (2019).
Haykin, Simon, “Adaptive Filter Theory”, Fourth Edition, Pearson Education, 2008.
Heathman, A. C. [1989] Methods for intermodulation prediction in communication systems. Ph. D. Thesis, University of Bradford, United Kingdom.
Heiskanen, Antti, Janne Aikio, and Timo Rahkonen. “A 5th order Volterra study of a 30W LDMOS power amplifier.” In Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS'03., vol. 4, pp. IV-IV. IEEE, 2003.
Helie, Thomas, “Introduction to Volterra series and applications to physical audio signal processing” IRCAM-CNRS UMR9912-UPMC, Paris, France DAFx, 2011.
Helie, Thomas, “Systemes entree-sortie non lineaires et applications en audio-acoustique”, Series de Volterra, Ecole Thematiqu “Theorie du Controlee n Mechanique” (2019).
Hermann, Robert. “Volterra modeling of digital magnetic saturation recording channels.” IEEE Transactions on Magnetics 26, No. 5 (1990): 2125-2127.
Hoerr, Ethan, and Robert C. Maher. “Using Volterra series modeling techniques to classify black-box audio effects.” In Audio Engineering Society Convention 147. Audio Engineering Society, 2019.
Ibnkahla, M., “Applications of neural networks to digital communications—a survey,” Signal processing, vol. 80, No. 7, pp. 1185-1215, 2000.
IESS Interlsat Earth Station Standards Document IESS-101 (Rev. 61) 2005.
IESS [Nov. 1996] IESS-401 (Rev. 4). Performance requirements for intermodulation products transmitted from INTELSAT earth stations. Intelsat Earth Station Standard (IESS).
Israelsen, Brett W., and Dale A. Smith. “Generalized Laguerre reduction of the Volterra kernel for practical identification of nonlinear dynamic systems.” arXiv preprint arXiv:1410.0741 (2014).
Kaeadar, K. [Dec. 1986] Gaussian white-noise generation for digital signal synthesis. IEEE Trans. Inst. and Meas., vol. IM 35, 4.
Kafadar, Karen. “Gaussian white-noise generation for digital signal synthesis.” IEEE transactions on instrumentation and measurement 4 (1986): 492-495.
Kahn, L. R. [Jul. 1952] SSB transmission by envelope elimination and restoration. Proc. IRE.
Kamiya, N., and F. Maehara. “Nonlinear Distortion Avoidance Employing Symbol-wise Transmit Power Control for OFDM Transmission,” Proc. of Int'l. OFDM Workshop, Hamburg, 2009.
Khan, A. A., and N. S. Vyas. “Application of Volterra and Wiener theories for nonlinear parameter estimation in a rotor-bearing system.” Nonlinear Dynamics 24, No. 3 (2001): 285-304.
Khan, A. A., and N. S. Vyas. “Non-linear parameter estimation using Volterra and Wiener theories.” Journal of Sound and Vibration 221, No. 5 (1999): 805-821.
Kim, J., and K. Konstantinou, “Digital predistortion of wideband signals based on power amplifier model with memory,” Electronics Letters, vol. 37, No. 23, pp. 1417-1418, 2001.
Koh, Taiho, and E. Powers. “Second-order Volterra filtering and its application to nonlinear system identification.” IEEE Transactions on acoustics, speech, and signal processing 33, No. 6 (1985): 1445-1455.
Kohli, Amit Kumar, and Amrita Rai. “Numeric variable forgetting factor RLS algorithm for second-order Volterra filtering.” Circuits, Systems, and Signal Processing 32, No. 1 (2013): 223-232.
Korenberg, Michael J., and Ian W. Hunter. “The identification of nonlinear biological systems: Volterra kernel approaches.” Annals of biomedical engineering 24, No. 2 (1996): 250-268.
Krall, Christoph, Klaus Witrisal, Geert Leus and Heinz Koeppl, “Minimum Mean-Square Error Equalization for Second-Order Volterra Systems”, IEEE Transactions on Signal Processing, vol. 56, No. 10, Oct. 2008.
Ku, Y. H., and Alfred A. Wolf. “Volterra-Wiener functionals for the analysis of nonlinear systems.” Journal of the Franklin Institute 281, No. 1 (1966): 9-26.
Leis, John, “Adaptive Filter Lecture Notes & Examples”, Nov. 1, 2008 www.usq.edu.au/users/leis/notes/sigproc/adfilt.pdf.
Li, B., C. Zhao, M. Sun, H. Zhang, Z. Zhou, and A. Nallanathan, “A bayesian approach for nonlinear equalization and signal detection in millimeter-wave communications,” IEEE Transactions on Wireless Communications, vol. 14, No. 7, pp. 3794-3809, 2015.
Li, Jian, and Jacek Ilow. “Adaptive Volterra predistorters for compensation of non-linear effects with memory in OFDM transmitters.” In 4th Annual Communication Networks and Services Research Conference (CNSR'06), pp. 4-pp. IEEE, 2006.
Liu, T., S. Boumaiza, and F. M. Ghannouchi, “Dynamic behavioral modeling of 3g power amplifiers using real-valued time-delay neural networks,” IEEE Transactions on Microwave Theory and Techniques, vol. 52, No. 3, pp. 1025-1033, 2004.
Lopez-Bueno, David, Teng Wang, Pere L. Gilabert, and Gabriel Montoro. “Amping up, saving power: Digital predistortion linearization strategies for power amplifiers under wideband 4GV5G burst-like waveform operation.” IEEE Microwave Magazine 17, No. 1 (2015): 79-87.
Álvarez-López, Luis, and Juan A. Becerra. “Application of deep learning methods to the mitigation of nonlinear effects in communication systems.”
López-Valcarce, Roberto, and Soura Dasgupta, “Second-Order Statistical Properties of Nonlinearly Distorted Phase-Shift Keyed (PSK) Signals”, IEEE Communications Letters, vol. 7, No. 7, Jul. 2003.
Lozhkin, Alexander N. “Turbo Linearizer for High Power Amplifier.” In 2011 IEEE 73rd Vehicular Technology Conference (VTC Spring), pp. 1-5. IEEE, 2011.
Lucciardi, J.-A., P. Potier, G. Buscarlet, F. Barrami, and G. Mesnager, “Non-linearized amplifier and advanced mitigation techniques: Dvbs-2x spectral efficiency improvement,” in GLOBECOM 2017-2017 IEEE Global Communications Conference. IEEE, 2017, pp. 1-7.
Maass, Wolfgang, and Eduardo D. Sontag. “Neural systems as nonlinear filters.” Neural computation 12, No. 8 (2000): 1743-1772.
Marmarelis, Vasilis Z., and Xiao Zhao. “Volterra models and three-layer perceptrons.” IEEE Transactions on Neural Networks 8, No. 6 (1997): 1421-1433.
Mathews, V. John, “Adaptive Polynomial Filters,” IEEE Signal Processing Magazine, vol. 8, No. 3, Jul. 1991.
Mathews, V. John. “Orthogonalization of correlated Gaussian signals for Volterra system identification.” IEEE Signal Processing Letters 2, No. 10 (1995): 188-190.
Mirri, Domenico, G. Luculano, Fabio Filicori, Gaetano Pasini, Giorgio Vannini, and G. P. Gabriella. “A modified Volterra series approach for nonlinear dynamic systems modeling.” IEEE Transactions on Circuits and Systems I: Fundamental Theory and Applications 49, No. 8 (2002): 1118-1128.
Mkadem, F., and S. Boumaiza, “Physically inspired neural network model for rf power amplifier behavioral modeling and digital predistortion,” IEEE Transactions on Microwave Theory and Techniques, vol. 59, No. 4, pp. 913-923, 2011.
Mollén, C., E. G. Larsson, and T. Eriksson, “Waveforms for the massive mimo downlink: Amplifier efficiency, distortion, and performance,” IEEE Transactions on Communications, vol. 64, No. 12, pp. 5050-5063, 2016.
Mzyk, Grzegorz, Zygmunt Hasiewicz, and Paweł Mielcarek. “Kernel Identification of Non-Linear Systems with General Structure.” Algorithms 13, No. 12 (2020): 328.
Niknejad, Ali M., “Volterra/Wiener Representation of Non-Linear Systems”, U.C. Berkeley (2014).
Nowak, Robert D., and Barry D. Van Veen. “Volterra filter equalization: A fixed point approach.” IEEE Transactions on signal processing 45, No. 2 (1997): 377-388.
Nowak, Robert D., and Barry D. Van Veen. “Random and pseudorandom inputs for Volterra filter identification.” IEEE Transactions on Signal Processing 42, No. 8 (1994): 2124-2135.
Ogunfunmi, Tokunbo. Adaptive nonlinear system identification: The Volterra and Wiener model approaches. Springer Science & Business Media, 2007.
Orcioni, Simone. “Improving the approximation ability of Volterra series identified with a cross-correlation method.” Nonlinear Dynamics 78, No. 4 (2014): 2861-2869.
Orcioni, Simone, Alessandro Terenzi, Stefania Cecchi, Francesco Piazza, and Alberto Carini. “Identification of Volterra models of tube audio devices using multiple-variance method.” Journal of the Audio Engineering Society 66, No. 10 (2018): 823-838.
Orcioni, Simone, Massimiliano Pirani, Claudio Turchetti, and Massimo Conti. “Practical notes on two Volterra filter identification direct methods.” In 2002 IEEE International Symposium on Circuits and Systems. Proceedings (Cat. No. 02CH37353), vol. 3, pp. III-III. IEEE, 2002.
Palm, G., and T. Poggio. “The Volterra representation and the Wiener expansion: validity and pitfalls.” SIAM Journal on Applied Mathematics 33, No. 2 (1977): 195-216.
Park, D.-C., and T.-K. J. Jeong, “Complex-bilinear recurrent neural network for equalization of a digital satellite channel,” IEEE Transactions on Neural Networks, vol. 13, No. 3, pp. 711-725, 2002.
Perev, K. “Orthogonal Approximation of Volterra Series and Wiener G-functionals Descriptions for Nonlinear Systems.” Information Technologies and Control (2019), Online ISSN: 2367-5357, DOI: 10.7546/itc-2019-0003.
Petrovic, V. and Gosling, W. [May 10, 1979] Polar loop transmitter. Elec. Lett.
Pirani, Massimiliano, Simone Orcioni, and Claudio Turchetti. “Diagonal Kernel Point Estimation of th-Order Discrete Volterra-Wiener Systems.” EURASIP Journal on Advances in Signal Processing 2004, No. 12 (2004): 1-10.
Prazenica, Richard J., and Andrew J. Kurdila. “Multiwavelet constructions and Volterra kernel identification.” Nonlinear Dynamics 43, No. 3 (2006): 277-310.
Radiocommunications Agency [Apr. 1987] Code of practice for radio site engineering. MPT 1331. Radiocommunications Agency (RA), Flyde Microsystems Ltd. United Kingdom.
Rai, Amrita, and Amit Kumar Kohli. “Analysis of Adaptive Volterra Filters with LMS and RLS Algorithms.” AKGEC Journal of Technology 2, No. 1 (2011).
Rawat, M., K. Rawat, and F. M. Ghannouchi, “Adaptive digital predistortion of wireless power amplifiers/transmitters using dynamic realvalued focused time-delay line neural networks,” IEEE Transactions on Microwave Theory and Techniques, vol. 58, No. 1, pp. 95-104, 2010.
Roheda, Siddharth, and Hamid Krim. “Conquering the CNN Over-Parameterization Dilemma: A Volterra Filtering Approach for Action Recognition.” In Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, No. 07, p. 11948-11956. 2020.
Root, D. E., J. Xu, T. Nielsen, C. Gillease, R. Biernacki, and J. Verspecht. “Measurement-based Nonlinear Behavioral Modeling of Transistors and Components.”
Rugh, Wilson J., Nonlinear System Theory: The Volterra/Wiener Approach. Johns Hopkins University Press, 1981.
Rupp, Markus, “Pre-Distortion Algorithm For Power Amplifiers”, Ph.D. Dissertation, eingereicht an der Technischen Universitat Wien.
Saadoon, Mohammed Ayad. “Linearization Of Power Amplifiers With Memory Effects Using Memory Polynomial And Linearly Interpolated Look-Up Table Predistorter.” (2017).
Saleh, A. M. [May 1982] Intermodulation analysis of FDMA satellite systems employing compensated and uncompensated TWT's'. IEEE Trans. Comm., vol. COM-30, 5.
Sáez, Raúl Gracia, and Nicolás Medrano Marqués. “RF power amplifier linearization in professional mobile radio communications using artificial neural networks.” IEEE Transactions on Industrial Electronics 66, No. 4 (2018): 3060-3070.
Saleh, Adel AM. “Intermodulation analysis of FDMA satellite systems employing compensated and uncompensated TWT's.” IEEE Transactions on Communications 30, No. 5 (1982): 1233-1242.
Sarti, Augusto, and Silvano Pupolin. “Recursive techniques for the synthesis of a pth-order inverse of a volterra system.” European Transactions on Telecommunications 3, No. 4 (1992): 315-322.
Schetzen, M., 1980, The Volterra and Wiener Theorems of Nonlinear Systems, New York, Chichester, Brisbane and Toronto: John Wiley and Sons.
Schetzen, Martin. “Synthesis of a class of non-linear systems.” International Journal of Control 1, No. 5 (1965): 401-414.
Schetzen, Martin. “Nonlinear system modeling based on the Wiener theory.” Proceedings of the IEEE 69, No. 12 (1981): 1557-1573.
Schetzen, Martin. “Multilinear theory of nonlinear networks.” Journal of the Franklin Institute 320, No. 5 (1985): 221-247.
Schetzen, Martin. “Nonlinear system modelling and analysis from the Volterra and Wiener perspective.” In Block-oriented Nonlinear System Identification, pp. 13-24. Springer, London, 2010.
Sebald, D. J., and J. A. Bucklew, “Support vector machine techniques for nonlinear equalization,” IEEE Transactions on Signal Processing, vol. 48, No. 11, pp. 3217-3226, 2000.
Shafi, Fahim. “Prediction of harmonic and intermodulation performance of frequency dependent nonlinear circuits and systems.” PhD diss., King Fahd University of Petroleum and Minerals, 1994.
Shimbo, O. [Feb. 1971] Effects of intermodulation, AM-PM conversion, and additive noise in multicarrier TWT systems. Proc. IEEE, vol. 59, p. 230-238.
Silva, Walter. “Reduced-order models based on linear and nonlinear aerodynamic impulse responses.” In 40th Structures, Structural Dynamics, and Materials Conference and Exhibit, p. 1262. 1999.
Sim, M. S., M. Chung, D. Kim, J. Chung, D. K. Kim, and C.-B. Chae, “Nonlinear self-interference cancellation for full-duplex radios: From link-level and system-level performance perspectives,” IEEE Communications Magazine, vol. 55, No. 9, pp. 158-167, 2017.
Simons, K., Technical Handbook for CATV Systems, 3rd Edition. Jerrod Publication No. 436-001-01, 1968.
Smith, C. N. Application of the Polar Loop Technique to UHF SSB transmitters. University of Bath (United Kingdom), 1986.
Sonnenschein, M & Censor, D., 1998, Simulation of Hamiltonian light beam propagation in nonlinear media, JOSA—Journal of the Optical Society of America B, (15): 1335-1345.
Staudinger, J., J.-C. Nanan, and J. Wood, “Memory fading volterra series model for high power infrastructure amplifiers,” in Radio and Wireless Symposium (RWS), 2010 IEEE. IEEE, 2010, pp. 184-187.
Stenger, Alexander, and Rudolf Rabenstein. “Adaptive Volterra filters for nonlinear acoustic echo cancellation.” In NSIP, pp. 679-683. 1999.
Stenger, Alexander, Lutz Trautmann, and Rudolf Rabenstein. “Nonlinear acoustic echo cancellation with 2nd order adaptive Volterra filters.” In 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No. 99CH36258), vol. 2, pp. 877-880. IEEE, 1999.
Stepashko, Volodymyr, 2nd International Conference on Inductive Modelling (ICIM'2008) Kyiv, Sep. 15-19, 2008.
Swiss Federal Institute of Technology, Lausanne, “Advantages of nonlinear polynomial predictors” (2010).
Tang, Hao, Y. H. Liao, J. Y. Cao, and Hang Xie. “Fault diagnosis approach based on Volterra models.” Mechanical Systems and Signal Processing 24, No. 4 (2010): 1099-1113.
Thapar, HK, and BJ Leon. “Lumped nonlinear system analysis with Volterra series[Final Phase Report, Nov. 1978-Nov. 1979].” (1980).
Therrien, Charles W., W. Kenneth Jenkins, and Xiaohui Li, “Optimizing the Performance of Polynomial Adaptive Filters: Making Quadratic Filters Converge Like Linear Filters”, IEEE Transactions on Signal Processing, vol. 47, No. 4, Apr. 1999.
Thomas, Erik GF, J. Leo van Hemmen, and Werner M. Kistler. “Calculation of Volterra kernels for solutions of nonlinear differential equations.” SIAM Journal on Applied Mathematics 61, No. 1 (2000): 1-21.
Tsimbinos John, and Langford B. White, “Error Propagation and Recovery in Decision-Feedback Equalizers for Nonlinear Channels”, IEEE Transactions on Communications, vol. 49, No. 2, Feb. 2001.
Tsimbinos, John, and Kenneth V. Lever. “Applications of higher-order statistics to modelling, identification and cancellation of nonlinear distortion in high-speed samplers and analogue-to-digital converters using the Volterra and Wiener models.” In [1993 Proceedings] IEEE Signal Processing Workshop on Higher-Order Statistics, pp. 379-383. IEEE, 1993.
Tsimbinos, John, and Langford B. White. “Error propagation and recovery in decision-feedback equalizers for nonlinear channels.” IEEE Transactions on Communications 49, No. 2 (2001): 239-242.
Tsimbinos, John, and Kenneth V. Lever. “The computational complexity of nonlinear compensators based on the Volterra inverse.” In Proceedings of 8th Workshop on Statistical Signal and Array Processing, pp. 387-390. IEEE, 1996.
Tsimbinos, John, and Kenneth V. Lever. “Computational complexity of Volterra based nonlinear compensators.” Electronics Letters 32, No. 9 (1996): 852-854.
Uncini, A., L. Vecci, P. Campolucci, and F. Piazza, “Complex-valued neural networks with adaptive spline activation function for digital-radio-links nonlinear equalization,” IEEE Transactions on Signal Processing, vol. 47, No. 2, pp. 505-514, 1999.
Vazquez, Rafael, and Miroslav Krstic. “Volterra boundary control laws for a class of nonlinear parabolic partial differential equations.” IFAC Proceedings vols. 37, No. 13 (2004): 1253-1258.
Wang, C.-L. and Y. Ouyang, “Low-complexity selected mapping schemes for peak-to-average power ratio reduction in ofdm systems,” IEEE Transactions on signal processing, vol. 53, No. 12, pp. 4652-4660, 2005.
Wang, Chin-Liang, and Yuan Ouyang. “Low-complexity selected mapping schemes for peak-to-average power ratio reduction in OFDM systems.” IEEE Transactions on signal processing 53, No. 12 (2005): 4652-4660.
Watanabe, Atsushi. “The volterra series expansion of functionals defined on the finite-dimensional vector space and its application to saving of computational effort for volterra kernels.” Electronics and Communications in Japan (Part I: Communications) 69, No. 4 (1986): 37-46.
Wei, Wentao, Peng Ye, Jinpeng Song, Hao Zeng, Jian Gao, and Yu Zhao. “A behavioral dynamic nonlinear model for time-interleaved ADC based on Volterra series.” IEEE Access 7 (2019): 41860-41873.
Woo, Young Yun, et al. “Adaptive Digital Feedback Predistortion Technique for Linearizing Power Amplifiers,” IEEE Transactions on Microwave Theory and Techniques, vol. 55, No. 5, May 2007.
Wood, J., Behavioral modeling and linearization of RF power amplifiers. Artech House, 2014.
Wray, Jonathan, and Gary GR Green. “Calculation of the Volterra kernels of non-linear dynamic systems using an artificial neural network.” Biological cybernetics 71, No. 3 (1994): 187-195.
Yan, H., and D. Cabric, “Digital predistortion for hybrid precoding architecture in millimeter-wave massive mimo systems,” in Acoustics, Speech and Signal Processing (ICASSP), 2017 IEEE International Conference on. IEEE, 2017, pp. 3479-3483.
Yasui, Syozo. “Stochastic functional Fourier series, Volterra series, and nonlinear systems analysis.” IEEE Transactions on Automatic Control 24, No. 2 (1979): 230-242.
Yoffe, I., and D. Wulich, “Predistorter for mimo system with nonlinear power amplifiers,” IEEE Transactions on Communications, vol. 65, No. 8, pp. 3288-3301, 2017.
Yu, Chao, Lei Guan, Erni Zhu, and Anding Zhu. “Band-limited Volterra series-based digital predistortion for wideband RF power amplifiers.” IEEE Transactions on Microwave Theory and Techniques 60, No. 12 (2012): 4198-4208.
Zaknich, A., “Principals of Adaptive Filter and Self Learning Systems”, Springer Link 2005.
Zhang, Haitao. On the design techniques to improve self-heating, linearity, efficiency and mismatch protection in broadband HBT power amplifiers. University of California, Irvine, 2006.
Zhu, Anding. “Behavioral modeling for digital predistortion of RF power amplifiers: from Volterra series to CPWL functions.” In 2016 IEEE Topical Conference on Power Amplifiers for Wireless and Radio Applications (PAWR), pp. 1-4. IEEE, 2016.
Zhu, A., J. C. Pedro, and T. J. Brazil, “Dynamic deviation reduction-based volterra behavioral modeling of rf power amplifiers,” IEEE Transactions on microwave theory and techniques, vol. 54, No. 12, pp. 4323-4332, 2006.
Zhu, A., P. J. Draxler, J. J. Yan, T. J. Brazil, D. F. Kimball, and P. M. Asbeck, “Open-loop digital predistorter for rf power amplifiers using dynamic deviation reduction-based volterra series,” IEEE Transactions on Microwave Theory and Techniques, vol. 56, No. 7, pp. 1524-1534, 2008.
Zhu, Anding, and Thomas J. Brazil. “Behavioral modeling of RF power amplifiers based on pruned Volterra series.” IEEE Microwave and Wireless components letters 14, No. 12 (2004): 563-565.
Zhu, Anding, and Thomas J. Brazil. “An adaptive Volterra predistorter for the linearization of RF high power amplifiers.” In 2002 IEEE MTT-S International Microwave Symposium Digest (Cat. No. 02CH37278), vol. 1, pp. 461-464. IEEE, 2002.
Zhu, Anding, Michael Wren, and Thomas J. Brazil. “An efficient Volterra-based behavioral model for wideband RF power amplifiers.” In IEEE MTT-S International Microwave Symposium Digest, 2003, vol. 2, pp. 787-790. IEEE, 2003.
Zhu, Anding, and Thomas J. Brazil. “RF power amplifier behavioral modeling using Volterra expansion with Laguerre functions.” In IEEE MTT-S International Microwave Symposium Digest, 2005., pp. 4-pp. IEEE, 2005.
Zhu, Anding, Paul J. Draxler, Jonmei J. Yan, Thomas J. Brazil, Donald F. Kimball, and Peter M. Asbeck. “Open-loop digital predistorter for RF power amplifiers using dynamic deviation reduction-based Volterra series.” IEEE Transactions on Microwave Theory and Techniques 56, No. 7 (2008): 1524-1534.
Zhu, Anding, Jos Carlos Pedro, and Telmo Reis Cunha. “Pruning the Volterra series for behavioral modeling of power amplifiers using physical knowledge.” IEEE Transactions on Microwave Theory and Techniques 55, No. 5 (2007): 813-821.
Zhu, Anding, Paul J. Draxler, Chin Hsia, Thomas J. Brazil, Donald F. Kimball, and Peter M. Asbeck. “Digital predistortion for envelope-tracking power amplifiers using decomposed piecewise Volterra series.” IEEE Transactions on Microwave Theory and Techniques 56, No. 10 (2008): 2237-2247.
Zhu, Anding, and Thomas J. Brazil. “An overview of Volterra series based behavioral modeling of RF/microwave power amplifiers.” In 2006 IEEE annual wireless and microwave technology conference, pp. 1-5. IEEE, 2006.
Zhu, Yun-Peng, and Zi-Qiang Lang. “A new convergence analysis for the Volterra series representation of nonlinear systems.” Automatica 111 (2020): 108599.

Related Publications (1)

	Number	Date	Country
	20230021633 A1	Jan 2023	US

Provisional Applications (1)

	Number	Date	Country
	62819054	Mar 2019	US

Continuations (2)

	Number	Date	Country
Parent	17234102	Apr 2021	US
Child	17947577		US
Parent	16812229	Mar 2020	US
Child	17234102		US

Integrating volterra series model and deep neural networks to equalize nonlinear power amplifiers

Information

Patent Number

Date Filed

Date Issued

Inventors

Original Assignees

Examiners

Agents

CPC

Field of Search

US

CPC

International Classifications

Abstract