Patent Application
20170250729
The present invention relates to wireless communication systems, including those that are frequency-hopped, that are subject to interference, and in particular, the estimation and mitigation of swept-tone interferers that are routinely employed as jamming signals.
Swept-tone interferers are usually characterized by linear frequency modulations: the instantaneous frequency of the signal sweeps a range of several megahertz (MHz) in a few microseconds, thereby affecting the entire band targeted by the interfering signal. A narrowband system or one or more hops of a frequency-hopped system, which is subjected to this broadband swept-tone interferer, may experience degraded system performance. Furthermore, the fast variations of the instantaneous frequency of swept-tone jammers make the design of estimation and mitigation techniques particularly challenging.
In an effort to mitigate swept-tone interferers, previous approaches developed include adaptive notch filters that can track the instantaneous frequency of the swept tone (see, for example, “Tracking and Mitigating a Jamming Sign al with an Adaptive Notch Filter,” Daniele Borio et al., Inside GNSS, March/April 2014), joint estimation of the parameters of a chirp signal using a least-squares approach (see, for example, “Parameter Estimation of Chirp Signals,” Petar M. Djurić and Steven M. Kay, IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol. 38, No. 12, pp. 2118-2126, December 1990), and tone prediction using an adaptive filter structure to enable cancellation of periodic interference signals (see, for example, the disclosure in U.S. Pat. No. 6,434,233 B1).
In comparison to these previously developed approaches, some of which directly estimate the parameters of the interfering signal, some embodiments of the present invention estimate the period of the swept-tone interferer and models it, in the observation bandwidth, as a magnitude-periodic signal with a common pulse shape. Each periodic segment of the interfering signal may be different from any other segment in phase. In other embodiments of the present invention, the magnitude-periodic model is over the hop bandwidth of a frequency-hopped system. In yet other embodiments of the present invention, the model is over all the hops of a frequency-hopped system that are affected by the swept-tone interferer. The interference estimate may be subtracted from the received filtered composite signal to generate an interference-mitigated signal of interest.
Embodiments of the present invention are directed towards methods and systems for the estimation and mitigation of swept-tone interferers. For example, in one embodiment, a method for estimating a swept-tone interferer comprises: receiving a composite signal over an observation bandwidth, wherein the composite signal comprises a signal of interest and the swept-tone interferer; modeling the swept-tone interferer over the observation bandwidth as a magnitude-periodic signal, wherein the magnitude-periodic signal comprises a plurality of phase-rotated pulses, and wherein each of the plurality of phase-rotated pulses comprises a common pulse shape and a distinct phase rotation; generating an energy signal based on the composite signal; estimating a period of the magnitude-periodic signal based on the energy signal; generating an initial estimate of the common pulse shape based on the period and energy detection of the composite signal, wherein a length of the initial estimate of the common pulse shape is equal to the period; generating a plurality of phase estimates based on a plurality of epochs and the initial estimate of the common pulse shape, wherein the plurality of epochs corresponds to non-overlapping, contiguous segments of the composite signal, wherein a length of each of the plurality of epochs is equal to the period, wherein each of the plurality of epochs comprises the common pulse shape, and wherein each of the plurality of phase estimates is generated for each of the plurality of epochs; generating a plurality of phase de-rotated pulse estimates, wherein each of the plurality of phase de-rotated pulse estimates is based on each of the plurality of epochs and each of the plurality of phase estimates; computing a weighted average of the plurality of phase de-rotated pulse estimates to generate a final estimate of the common pulse shape; and generating an estimate of the swept-tone interferer based on the final estimate of the common pulse shape and the plurality of phase estimates. In some embodiments, the method for mitigating the swept-tone interferer further comprises: subtracting the estimate of the swept-tone interferer from the composite signal to generate an interference-mitigated signal of interest.
Other embodiments of the present invention are directed towards methods and systems for the estimation of swept-tone interferers in frequency-hopped systems. For example, in one embodiment, the frequency-hopping system comprises a plurality of hops, each of the plurality of hops comprising each of a plurality of hop bandwidths, a method for estimating a swept-tone interferer in a frequency-hopping system comprises: sampling a composite signal over each of the plurality of hop bandwidths to generate an initial set of samples for each of the plurality of hops, wherein the composite signal comprises a signal of interest and the swept-tone interferer, and wherein each of the initial set of samples comprises a first set of samples and a second set of samples; determining a number of the first set of samples for each of the plurality of hops; truncating the first set of samples from the initial set of samples for each of the plurality of hops; concatenating the truncated set of samples for each of the plurality of hops to generate a reconstructed composite signal; modeling the swept-tone interferer in the reconstructed composite signal as a magnitude-periodic signal, wherein the magnitude-periodic signal comprises a plurality of phase-rotated pulses, and wherein each of the plurality of phase-rotated pulses comprises a common pulse shape and a distinct phase rotation; estimating a period of the magnitude-period signal; generating an initial estimate of the common pulse shape based on at least the period, wherein a length of the initial estimate of the common pulse shape is equal to the period; generating a plurality of phase estimates based on a plurality of epochs and the initial estimate of the common pulse shape, wherein the plurality of epochs corresponds to non-overlapping, contiguous segments of the reconstructed composite signal, wherein a length of each of the plurality of epochs is equal to the period, wherein each of the plurality of epochs comprises the common pulse shape, and wherein each of the plurality of phase estimates is generated for each of the plurality of epochs; generating a plurality of phase de-rotated pulse estimates, wherein each of the plurality of phase de-rotated pulse estimates is based on each of the plurality of epochs and each of the plurality of phase estimates; computing a weighted average of the plurality of phase de-rotated pulse estimates to generate a final estimate of the common pulse shape; and generating an estimate of the swept-tone interferer based on the final estimate of the common pulse shape and the plurality of phase estimates.
These illustrative embodiments are mentioned not to limit or define the limits of the present subject matter, but to provide examples to aid in the understanding thereof. Illustrative embodiments are discussed in the Detailed Description, and further examples are provided there. Advantages offered by various embodiments may be further understood by examining this specification and/or by practicing one or more embodiments of the claimed subject matter.
Like labels are used to refer to the same or similar features in the drawings.
The present invention is well-suited for the estimation and mitigation of swept-tone interferers, i.e. jamming signals that are frequency-modulated with a fast varying center frequency. The time-frequency evolution, or instantaneous frequency as function of time, for representative swept-tone interferers are shown in
J(t)=A·cos(2πξ(t)t+θ), Eq (1)
where ξ(t) is the instantaneous frequency as a function of time, t ∈ (0, T), θ is a random initial phase, and 1/T is the repetition frequency of the sweep. In an embodiment, the instantaneous frequency is a linear function, which may be specified as:
In another embodiment, the instantaneous frequency may be any function of time that adheres to the edge conditions defined below for each epoch, k ∈ [0, K], in the received signal:
ξ(0)=fstart; ξ(kT−)=fstop; and ξ(kT+)=fstart. Eq (3)
In the example in
Embodiments of the present invention are also well-suited for the estimation and mitigation of swept-tone interferers in representative frequency-hopped systems, shown in
As noted earlier, the representative linear swept-tone interferers shown in
The swept-tone interferer enters the frequency band of the first hop 311 at a first delay 317 (denoted τ1) relative to the start of the hop at time t0. The interferer continues to sweep through the hop for the duration of the hop 323 (denoted Thop), and enters the second hop 313 at a second delay 321 (denoted τ2) relative to the start of the hop at time (t0+Thop). As is evident from
In an embodiment, if a delay for a current frequency hop (τi) is known, the delay for the next frequency hop(τi+1), in the case of a linear swept-tone interferer, can be determined using:
The relative delay due to the frequency separation Δf and the sweep rate α may be denoted Δτ. The relationship Δτ=Δf/α shown in Equation (4) is valid for swept-tone interferers with linear sweeps. However, for the more general swept-tone defined in Equation (1), the sweep rate will also be a function of time (α(t)), and the necessary parameters can be determined by the following relationship:
∫0Δτα(t)dt=Δf. Eq (5)
In another embodiment, if the delay for a first frequency hop (τi) is known, the delay for any other frequency hop (τj) can be estimated by cross-correlating the samples of the first frequency hop and the samples of the other frequency hop. The samples for each frequency hop include a first set of samples that correspond to a period of time prior to the interferer entering that frequency hop, and a second set of samples that correspond to the frequency hop being affected by the swept-tone interferer. Thus, cross-correlating the samples from the two frequency hops will have a first correlation peak at a lag value that equals (τi−τj).
In yet another embodiment, if the delay for a first frequency hop (τi) is known, the delay for other frequency hops can be estimated by cross-correlating the samples of the first frequency hop and the samples for multiple other frequency hops. For example, the samples for the first frequency hop may be correlated with the samples for the second and third frequency hops, and subsequently with the samples for both the third and fourth frequency hops. The first peaks of the correlation outputs may provide coarse delay values. This may be followed by a fine search around the coarse delay values to maximize a sum correlation score. Other embodiments may rely on the cross-correlations between sets of samples from different combinations of multiple frequency hops to provide the coarse delay values.
The waveforms shown in
v
k(t)=p(t)exp(jφk). Eq (6)
Consequently, the magnitude-periodic signal may be represented as
where T is the period of the magnitude-periodic signal and therefore also the length of each epoch. This signal representation is used to model the swept-tone interferer in the observation bandwidth. In another embodiment, each epoch may further comprise a distinct amplitude, i.e.
v
k(t)=Akp(t)exp(jφk). Eq (8)
In an embodiment, a composite signal comprising a signal of interest s and a swept-tone interferer J is received and filtered using a low-pass filter 605. As described in
A phase estimation module 655 uses the initial pulse estimate p0(t) in conjunction with the filtered composite signal to generate estimates for the set of distinct phase rotations {φk}k=1K for each of the epochs. The final pulse estimation module 675 generates a final estimate of the common pulse shape {circumflex over (p)}(t) based on the initial pulse estimate and the set of distinct phase rotations. This final pulse estimate of the common pulse shape, and the set of distinct phase rotations, is used by an interference estimate module 695 to generate an estimate of the swept-tone interferer. Summer 625 is used to subtract the estimate of the swept-tone interferer from the filtered composite signal to generate an interference-mitigated signal of interest.
The embodiment shown in
The filtered composite signals for each of the frequency hops are processed by a delay estimation module 707 that determines the delay for each set of samples. As discussed in the context of
In another embodiment, when the parameters of the swept-tone interferer are unknown, the set of samples corresponding to the first frequency hop of the packet/transmission may be used to compute an energy signal, and energy detection of that energy signal may be employed to determine the delay (τ0) for the first frequency hop. Delays for other frequency hops can be determined by cross-correlating the samples corresponding to the first frequency hop and the samples corresponding to another frequency hop, as described in the context of
The delays for each of the frequency hops are used by the signal conditioning module 709 to reconstruct the composite signal from sets of samples from each of the frequency hops. That is, the delay of a frequency hop corresponds to the number of the first set of samples prior to the interferer entering that particular frequency hop. The signal conditioning module 709 truncates the first set of samples from each set of samples for each frequency hop, and then concatenates the remaining (or second) set of samples from each frequency hop to generate a reconstructed composite signal. The second set of samples contains signal of interest samples that are subject to interference, and concatenating them reconstructs the composite signal.
In an embodiment, the signal conditioning module 709 may be used in the context of
In the reconstructed composite signal, the swept-tone interferer may be modeled as a magnitude-periodic signal, as discussed previously. In an embodiment, the pulse and phase estimation module 760 may be an amalgamation of the phase estimation module 655 with the final pulse estimation module 675, shown in
In an embodiment, the magnitude-square of the filtered composite signal (also referred to as an energy signal) is computed using a squaring module 812, and serves as an input to a fast Fourier transform (FFT) block 814 and an energy detector 818. The FFT block 814 computes a frequency spectrum of the energy signal, and the period estimator 816 uses at least one harmonic of the frequency spectrum to estimate the period T of the energy signal.
The initial estimate of the common pulse shape p0(t) is generated using a pulse generator 822, and is based on the period T and the output of the energy detector 818. In an embodiment, and in reference to
Demultiplexer 824 is used to partition the non-overlapping, contiguous epochs of the filtered composite signal into individual segments, denoted v1, v2, . . . , vK. Each of these epochs is correlated with the initial pulse estimate using a bank of K correlators 826.1, . . . , 826.K to generate estimates for the distinct phase rotations {φk}k=1K for each of these epochs, i.e.
{circumflex over (φ)}k=angle(v0, vk
), k=1, . . . , K, Eq (9)
where v0, vk
=∫0Tv*0(t)vk(t)dt is the inner product of the sampled epochs v0(t) and vk(t).
The estimated phases are used by a unit vector generator 828 to generate phase vectors with unit magnitudes and phases that are based on the estimated distinct phase rotations. In an embodiment, a first bank of K multipliers 832.1, . . . , 832.K is used to multiply each of the partitioned epochs by unit vectors with phases that are equal to the negative of the estimated distinct phase rotations. That is, each of the epochs v1, v2, . . . , vK with its distinct phase rotations is de-rotated using the estimates of {φk}k=1K. An averaging unit 834 generates a final estimate of the common pulse shape {circumflex over (p)}(t) by averaging the de-rotated epochs, i.e.
where {circumflex over (φ)}0=0, and {circumflex over (φ)}k is as specified in Equation (9).
In another embodiment, each of the epochs comprises a distinct amplitude (as described in Equation (8)); the amplitude for the kth epoch may be estimated as
and the corresponding final estimate of the common pulse shape is computed as
Note that the weighted average in Equation (12) simplifies to the final estimate of the common pulse shape in Equation (10) when all the amplitudes are equal to each other.
Demultiplexer 836 generates K+1 copies of the final estimate of the common pulse shape, each of which is multiplied by a unit vector with a phase equal to the respective estimated distinct phase rotation using a second bank of K+1 multipliers 838.0, . . . , 838.K. The distinctly rotated final pulse estimates are concatenated using a multiplexer 842 to generate an estimate of the swept-tone interferer, given by
In an embodiment, this estimate may be subtracted from the filtered complex-valued composite signal to generate an interference-mitigated signal of interest.
As shown in
The initial pulse estimation module 835 is realized using an energy detector and a pulse generator that generates an initial estimate of the common pulse shape by extracting the first epoch of the swept-tone interferer. In particular, the energy detector processes the energy signal to determine the start and end points of the common pulse shape in the first epoch, and the period is then used to extract the first epoch. In another embodiment, the autocorrelation of the magnitude-periodic signal may be used to determine the initial pulse estimate. For example, a segment that begins at a minimum value of the autocorrelation, and extends for the length of a period, may be used as the initial estimate of the common pulse shape.
In yet another embodiment, the initial pulse estimate may be correlated with each of the epochs in an effort to verify that a robust estimate of the common pulse shape has been extracted (as compared to a spurious feature with a comparable amount of energy) prior to the estimation of the distinct phase rotations. In an example, normalized correlation values between 0.85 and 1.0 are expected if a “correct” epoch has been extracted as the initial pulse estimate.
In this embodiment, the phase estimation module 855 is realized using a bank of correlators that correlate each of the epochs with the initial pulse estimate. As described in reference to
With reference to
At step 920, the swept-tone interferer over the observation bandwidth is modeled as a magnitude-periodic signal comprising a set of non-overlapping, contiguous phase-rotated pulses. In an embodiment, each of the phase-rotated pulses comprises a common pulse shape and a distinct phase rotation. In another embodiment, the phase-rotated pulses further comprise distinct amplitudes.
At step 930, an energy signal is computed based on the composite signal using a magnitude-squaring function. The energy signal is a periodic signal, and may be segmented into epochs, each of which comprises at least the common pulse shape.
At step 940, the period of the magnitude-periodic signal is estimated based on at least one harmonic of a frequency spectrum of the energy signal. In an embodiment, the fast Fourier transform (FFT) is used to compute the frequency spectrum, and the period is based on only the strongest (and typically fundamental) harmonic of the frequency spectrum. In another embodiment, multiple harmonics (at frequencies 1/T, 2/T, 3/T, . . . ) are used to estimate the period, via an average or a weighted average.
At step 950, an initial estimate of the common pulse shape is generated based on the period and the energy detection of the energy signal. In an embodiment, an energy detector is used to determine the start and end (in time) of the common pulse shape and knowledge of the period enables the generation of an initial estimate of the common pulse shape. For example, the first epoch may be used as an initial estimate of the common pulse shape.
At step 960, the set of distinct phase rotations is estimated based on a plurality of epochs and the initial estimate of the common pulse shape. As noted earlier, the epochs are non-overlapping, contiguous segments of the filtered composite signal, wherein each epoch comprises a common pulse shape and a distinct phase rotation (which is being estimated at this step). In an embodiment, the initial estimate of the common pulse shape is correlated with each of the plurality of epochs to generate the distinct phase rotation estimate.
At step 970, a plurality of phase de-rotated pulse estimates are generated by multiplying each of the epochs by a unit vector that has a phase that is equal to the negative of the estimated distinct phase rotation for that epoch. In another embodiment, when the magnitude-periodic signal is modeled with both a distinct phase and amplitude, the kth pulse estimate is given by Âkvk(t) exp(−j{circumflex over (φ)}k).
At step 980, a weighted average of the plurality of phase de-rotated pulse estimates is computed to generate a final estimate of the common pulse shape. As is expected, the final estimate of the common pulse shape of the swept-tone interferer will be minimally affected by AWGN and the signal of interest due to the averaging of all the epochs.
At step 990, the final estimate of the common pulse shape is rotated by each of the distinct phase rotations, and the rotated final estimates are concatenated to generate an estimate of the swept-tone interference. In an embodiment, the estimate of the swept-tone interference is subtracted from the filtered composite signal to generate an interference-mitigated signal of interest.
With reference to
In an embodiment, the composite signal is first processed through a bank of low-pass filters, where the passband of each of the low-pass filters corresponds to the frequency band for each of the frequency hops of the frequency-hopped system. That is, the center frequency of each of the low-pass filters in the filter bank corresponds to the center frequency of the frequency band for each of the frequency hops.
Sampling the composite signal, using an analog-to-digital converter, over the hop bandwidth for each of the frequency hops, generates a set of initial samples for each frequency hop. As discussed above in the context of
At step 1012, and as discussed above, the number of samples in the first set of samples for each initial set of samples is determined. This determination may be based on parameters of the swept-tone interferer, if they are known, or by energy detection on the initial samples of the first hop followed by cross-correlating the initial samples of the first hop with the initial samples of every other frequency hop. That is, the delay for each frequency hop is determined.
At step 1014, after having determined the number of the first set of samples for each initial set of samples, the first set of samples is truncated. At step 1016, the second set of samples for each frequency hop is concatenated, in the order of the frequency-hopping sequence, to generate a reconstructed composite signal.
The reconstructed composite signal, having been combined over all frequency hops, now contains a far greater number of sweeps of the interferer, and enables modeling the interferer as a magnitude-period signal, which is performed at step 1020. The modeling is similar to that described in step 920 in
At step 1040, the period of the magnitude-periodic signal is estimated. In the case where the parameters of the swept-tone interferer are known, the period may be estimated as the inverse of the repetition rate. If the parameters are unknown, the method according to steps 930 and 940 in
With reference to
As shown in flowchart 1100, steps 1120 through 1190 are performed for each of the N hops of a frequency-hopped system. They are functionally similar to steps 920 through 990 as shown in
In an embodiment, the system 1200 comprises member nodes 1220-1 . . . 1220-4 of a wireless network and a nearby radar installation 1210. The communication of the network nodes may be subject to interference due to the radar sweep signals, whose power is typically significantly higher than the power of the inter-node communications. In the observation bandwidth of the network nodes, the radar sweep signal acts as a swept-tone interferer that can be estimated and mitigation by embodiments of the present invention.
That is, each of the network nodes 1220-1 . . . 1220-4 may be configured as the system 1300 shown in
In an embodiment, the PIE unit comprises a processor 1304 and a memory 1303. The processor 1304 may comprise component digital processors, and may be configured to execute computer-executable program instructions stored in memory 1303. For example, the component digital processors may execute one or more computer programs for enabling the estimation and mitigation of swept-tone interferers in accordance with embodiments of the present invention.
Processor 1304 may comprise a variety of implementations for truncating and concatenating samples, estimating a period of the magnitude-periodic signal, an initial and final pulse estimate of common pulse estimate, a set of distinct phase rotations, and computing a weighted average of pulse estimates to generate an estimate of the swept-tone interferer, as well as a microprocessor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), one or more field programmable gate arrays (FPGAs), state machines, or the like. Processor 1304 may further comprise a programmable electronic device such as a programmable logic controller (PLC), a programmable interrupt controller (PIC), a programmable logic device (PLD), a programmable read-only memory (PROM), an electronically programmable read-only memory (EPROM or EEPROM), or another similar device.
Memory 1303 may comprise a non-transitory computer-readable medium that stores instructions which, when executed by the processor 1304, cause the processor 1304 to perform various steps, such as those described herein. Examples of computer-readable media include, but are not limited to, electronic, optical, magnetic, or other storage or transmission devices capable of providing the processor 1304 with computer-readable instructions. Other examples of computer-readable media comprise, but are not limited to, a floppy disk, CD-ROM, magnetic disk, memory chip, ROM, RAM, ASIC, configured processor, any optical medium, any magnetic tape or other magnetic medium, or any other medium from which a computer processor can access data. In addition, various other devices may include a computer-readable medium such as a router, private or public network, or other transmission device. The processor 1304 and the processing described may be in one or more structures, and may be dispersed throughout one or more structures.
Embodiments in accordance with aspects of the present subject matter can be implemented in digital electronic circuitry, computer hardware, firmware, software, or in combinations of the preceding. In one embodiment, a computer may comprise a processor or processors. A processor comprises or has access to a computer-readable medium, such as a random access memory (RAM) coupled to the processor.
While the present subject matter has been described in detail with respect to specific embodiments thereof, it will be appreciated that those skilled in the art, upon attaining an understanding of the foregoing, may readily produce modifications to, variations of, and equivalents to such embodiments. Accordingly, it should be understood that the present disclosure has been presented for purposes of example rather than limitation, and does not preclude inclusion of such modifications to, variations of and/or additions to the present subject matter as would be readily apparent to one of ordinary skill in the art.
This application is a continuation-in-part of U.S. application Ser. No. 15/053,966, filed Feb. 25, 2016, which is hereby expressly incorporated by reference in its entirety for all purposes.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15053966 | Feb 2016 | US |
| Child | 15374285 | US |
