The present disclosure relates generally to methods for filtering, and, more particularly, to a method and apparatus for mitigation of GNSS-signal interference using an adaptive notch filter.
The transmission of signals is typically subject to interference from multiple sources. Sources of interference can be natural or man-made and may also be intentional or unintentional. Interference and noise can prevent operations that require receipt of a useful signal. For example, thermal noise or interference jamming (e.g., signal blocking), can cause signal loss and prevent operations that require receipt of the transmitted signal. Global navigation satellite system (GNSS) signals from the satellites of those systems are typically weak in strength and, therefore, susceptible to interference. What is needed is a method to mitigate signal interference.
In one embodiment, an apparatus comprises a notch filter having a tunable frequency of a transfer function zero (also referred to as the “zero frequency”) and configured to receive an input signal and generate an output signal. A bandpass filter is coupled to an output of the notch filter and configured to receive the output signal. An adaptive block is coupled to the bandpass filter and configured to adjust the notch filter parameters in order to minimize a specific cost function. In one embodiment, the notch filter is a digital complex filter of a 1st order, wherein the input of the digital complex filter is coupled to an input of the apparatus and the frequency of the transfer function zero of the digital complex filter is equal to an interference frequency when adaptation is complete. In one embodiment, the bandpass filter is a digital complex filter of a 1st order having a pole frequency that coincides with the zero frequency of the notch filter. The digital complex filter has a bandpass filter transfer function, and the input of the digital complex filter is coupled to an output of the apparatus. In one embodiment, the adaptive block is configured to track an interference frequency and adjust the filter zero in order to achieve minimization of a cost function, the input of the adaptive block is coupled to the output of the bandpass filter. In one embodiment, the apparatus is configured to mitigate multi-spectral interference with the notch filter having a particular transfer function. In one embodiment, a highpass filter is shifted using real coefficients by multiplying the real coefficients by a power function of the complex exponent. A method is also described having the step of receiving an input signal from one or more global navigation satellite system satellites. The input signal is filtered by a notch filter and input to a bandpass filter, the output signal of which is input to the adaptive block, which adjusts the notch filter parameters in accordance with a Least Mean Squares (LMS) algorithm. An apparatus comprising a processor and a memory coupled to the processor is also described. The memory storing computer program instructions that when executed cause the processor to perform operations.
A method and apparatus for mitigation of GNSS-signal interference using an adaptive notch filter (ANF) operates based on signals received from one or more satellites of a Global Navigation Satellite System (GNSS) such as GPS, GLONASS, etc.
where, x[n]=Σkxk[n] is a combination of GNSS signals from different satellites, i[n]—is an interference signal, and η[n]—is thermal noise with spectral density N0.
GNSS signal interference mitigation in the time domain can involve the removal of the interference signal i[n]from the signal y[n]:
Narrowband quasi-harmonic interference and a chirp signal can be represented in the following single-component form:
In this scenario, the instantaneous value of the interference signal frequency (equation 3) has the form:
In the case of a single-component signal equation (3), a prediction regarding the form of the interference signal i[n] from the previous sample i[n−1]can be used as follows:
The expression for the coefficient a[n] can be obtained from equations (3) and (5) taking into account that, for neighboring samples, if the noise amplitude is considered to not change, then the coefficient of linear prediction of the error signal is determined by estimating only the instantaneous value of the interference frequency fi:
In one embodiment, an interference mitigation system for a GNSS receiver uses a digital notch filter with complex transfer function of 1st order (also referred to as a digital complex filter of a 1st order) with a transfer function:
where z0[n]—is complex zero of the transfer function (equation (7)), and kα<1—is the real coefficient at the pole of the filter's transfer function.
The transfer function numerator of equation (7) is referred to as the Moving Average (MA) part. MA components 106 shown in
From equation (8), it follows that a single-component interference signal is mitigated when the interference signal estimate (equation (6)) coincides with the zero value of the notch filter tuned to the interference with the filter transfer function (equation (7)) which can be shown as:
The autoregressive (AR) part of the filter transfer function
reduces the MA influence on the distortion of the desired (i.e., useful) signal by narrowing the rejection bandwidth of the filter. AR components 108 shown in
In accordance with the Least Mean Squares (LMS) algorithm, the search for the optimal value (also referred to as the target value) of the zero frequency of the notch filter is performed iteratively at each sample. The direction of the search and the value of the corrective additive itself is related to the value of the gradient of the cost function, which is equal to the power of the output signal:
where xf[n]—is the output signal of the MA block in the notch filter. With successful adaptation, i.e. reaching the minimum of the cost function, the zero frequency of the notch filter coincides with the interference frequency with a given accuracy.
The limitation in increasing the interference mitigation depth is the reduction in the ratio of the interference signal power at the output of the notch filter J to the noise power
where N0—is the thermal noise spectral density, and B—is the bandwidth of useful GNSS signal.
For a two-sided spectrum of a digital signal with the Nyquist frequency fs/2, bandwidth frequency is B=2·(fs/2)=fs. In the case of a narrowband interference signal, it is not generally possible to achieve a reduction in the spectral density of the interference signal to the level of the spectral density of thermal noise due to the much smaller bandwidth of the interference signal ΔΩi compared to the bandwidth of the noise:
It should be noted that bandpass filter 206 located at the input of adaptive block 204 does not change the signal in the direct transmission channel, since a bandpass filter is used in a computational circuit of the adaptive notch filter 200 of the GNSS receiver.
The width of the rejection/stop band of notch filter 202 with the transfer function of the form shown in equation (7) is B−3 dB=(1−kα)fsπ/10. In one embodiment, it is assumed that the stop band of notch filter 202 coincides with the frequency band of the interference signal (B−3 dB=ΔΩi). When the bandwidth of notch filter 202 is related to the bandwidth of the interference signal, then reduction in noise level at the input of adaptive block 204 and a gain in the ratio J/N is
A bandpass filter at the input of the adaptive block should be tunable because, during the adaptation of the notch filter, its pole-frequency changes, also tuning to the frequency of the interference signal. But the adjustment of the digital filter requires recalculation of the transfer function coefficients in real-time. For example, in one embodiment, a biquad filter is used as a bandpass filter with a resonant frequency Ωi=2πfi and Q-factor Q. After bilinear frequency transformation
of the biquad, which is an analog embodiment, the transfer function of the digital biquad in the z-domain has the following form:
To adjust the resonant frequency of a digital filter with transfer function (equation (11)), the coefficients a1 and a2 are changed. In one embodiment, the zero frequency of a transfer function of the digital complex filter is equal to the interference frequency when adaptation is complete.
In addition, after frequency conversion, distortions in the frequency characteristics occur, the level of which increases as the resonant frequency approaches the Nyquist frequency. There is a shift in the value of the resonant frequency from the calculated value. To accurately implement a digital filter at the frequency Ω, predistortion (see equation (8)) is introduced according to
In this case coefficient β in expression (11) should be in the form:
As a result, when tuning the zero frequency of the notch filter in the adaptation process, it is required to calculate the coefficients of the transfer function of the digital biquad filter with the sampling frequency according to the following equations:
Equations (12) require relatively large computational resources when implementing interference mitigation with a Field Programmable Gate Array (FPGA).
The transfer function with real coefficients (see equation (11)), in the general case, has complex conjugate zeros and poles. For a two-sided spectrum, the frequency response of a filter with real coefficients has a symmetrical response, both in the positive frequency region and in the conjugate negative part. But the interference signal, as a rule, is located in one of the two conjugate parts of the spectrum. Then the bandpass filter with a real transfer function has an excess bandwidth in the conjugate part of the spectrum, where there is no interference signal, which at least impair the value of the ratio J/N by 3 dB.
In one embodiment, bandpass filter 406 operates as a transfer function, specifically:
In one embodiment, bandpass filter 406 has the following characteristics:
The transfer function shown in equation (13) corresponds to a digital complex 1st order bandpass filter with a complex pole z=kbz0=kbejΩ
is zero frequency of the transfer function of notch filter 402 (see equation (7)), i.e., the value of the resonant frequency of the proposed bandpass filter coincides with the zero frequency of the transfer function of notch filter 402.
The real coefficient kb<1 determines the bandwidth of the bandpass filter, the value of which can also be estimated by the formula B−3 dB=(1−kb)fSπ/10. The closer the value of the kb coefficient is to one, the smaller the bandwidth of the given band pass filter.
The transfer coefficient of the proposed filter at the resonant frequency z=z0=ejΩ
Finally, the bandpass complex filter operating in accordance with the transfer function shown in equation (13) guarantees the selectivity only in one (of two) conjugate parts of two-sided frequency spectrum.
In one embodiment, the bandpass filter comprises a digital complex 1st order filter having a pole frequency that coincides with the zero frequency of the notch filter, the digital complex filter having a transfer function of
Methods for mitigating interference are generally limited to mitigating only one type of interference. For example, the PB (pulse blanker) method works satisfactorily with wideband pulsed interference signals, but is not designed to deal with continuous interference. Multi-spectral methods using discrete Fourier transformation (DFT), discrete Wavelet Transform (DWT) or Karhunen-Loève transform (KLT) provide great flexibility for interference mitigation, but require very large computational resources.
The adaptive notch filtering (ANF) method described herein provides advantageous results in the case of quasi-harmonic interference and chirp signals, but may be considered less effective for mitigation of multi-spectral or impulse interference signals. However, the effectiveness of the adaptive notch filtering method can be increased as follows.
In one embodiment, a notch-type digital complex 1st order bandpass filter operates according to the transfer function of equation (7). The digital complex 1st order bandpass filter described herein does not require additional computing resources to recalculate the transfer function and provides tracking and mitigating of chirp-interference with a frequency rate up to 10 MHz/μs. In one embodiment, the adaptive block is configured to track an interference frequency and adjust the filter zero in order to achieve minimization of a cost function. In operation of the digital complex 1st order filter, the real coefficient kαsimultaneously affects the signal mitigation depth and the rejection bandwidth. To mitigate broadband (or, for example, multi-spectral type OFDM signal) interference, the rejection band of the notch filter is increased, i.e. kα is diminished. But in this case, the degree of signal mitigation in the stopband is reduced. To resolve this contradiction, the filter order is increased and the stopband of the notch filter is expanded. One of two possible solutions may be implemented to resolve the contradiction.
In the first approach, a digital complex 3rd order filter is used consisting of three cascaded notch filters with the resulting transfer function in the form:
It should be noted that when changing (i.e., tuning) during adaptation, the zero frequency of the notch filter f0, the value Δf1, which determines the rejection band, does not need to be changed. The value of the complex poles of these additional two 1st order filters z0±Δ, does not require complicated calculations, it is determined by the sum of the arguments of two complex numbers.
In general, for multi-spectral interference with a larger bandwidth, it is required to increase the order of the notch filter. In addition, the location of zeros in the approximation function must be symmetrical about the central frequency. The transfer function of such a filter can be represented as:
The number of zeroes and their location (Δf1<Δf2< . . . <Δfk) is determined by the required depth of interference mitigation in the stopband of the 2(k+1)-th order notch filter. The frequency value f0 must correspond to the center frequency of the interference spectrum, and the value 2·Δfk—to the signal width of this interference.
The second approach implements a notch-type high order complex filter for the interference mitigation by shifting the highpass filter with real coefficients by multiplying the real coefficients by a power function of the complex exponent. Either a Finite Impulse Response (FIR) filter with real coefficients of pulse response h(m): yhp(n)=Σm=0M
can be used as a highpass filter. The highpass filter with real coefficients provides signal attenuation for the two-sided spectrum within band [−Ωc, +Ωc].
The coefficients of the complex filter are further formed by multiplying the coefficients of the original highpass filter by a power function of the complex exponent: h(m)·e−jΩ
As a result, the transfer function of the complex IIR filter is obtained
or the pulse response characteristic of the complex FIR-filter ynotch(n)=Σm=0M
e−jΩ
In one embodiment, the method can be used to detect and mitigate multi-spectral interference (type of orthogonal frequency division multiplexing (OFDM) signal) when using high-order notch filter in ANF structure.
The approach and techniques described herein provides detection and
mitigation of not only single-component interference (CW and chirp), but also multi-spectral interference, which gives a qualitative advantage over other works devoted to ANF-based interference mitigation systems.
The techniques described herein can be used for detecting and mitigating not only single-component interference (like CW and chirp), but also for combating multi-spectral wideband interference.
In one embodiment, a computer is used to perform the operations of the components and equations described herein and shown, for example, in
The foregoing Detailed Description is to be understood as being in every respect illustrative and exemplary, but not restrictive, and the scope of the inventive concept disclosed herein should be interpreted according to the full breadth permitted by the patent laws. It is to be understood that the embodiments shown and described herein are only illustrative of the principles of the inventive concept and that various modifications may be implemented by those skilled in the art without departing from the scope and spirit of the inventive concept. Those skilled in the art could implement various other feature combinations without departing from the scope and spirit of the inventive concept.
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/RU2022/000171 | 5/23/2022 | WO |