A Method for Joint Impulse Interference and Narrowband Interference Cancellation in a Time-Varying OFDM System

Abstract

The present invention belongs to the field of wireless communication technologies, and particularly relates to a method for jointly eliminating impulse interference and narrowband interference in a time-varying OFDM system. Aiming at the carrier frequency offset and joint interference existing in the time-varying OFDM system, the present invention first compensates the carrier frequency offset by using a preamble, then estimates and eliminates the narrowband interference according to the large amplitude characteristic of the narrowband interference in the frequency domain, and finally estimates and eliminates the impulse interference by measuring null subcarriers and utilizing the time-domain sparsity of the impulse interference. The present invention is applicable to a mobile communication environment with joint interference, and the proposed method can achieve frequency offset compensation and interference elimination, thereby improving the communication quality.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and particularly relates to a method for jointly eliminating impulse interference and narrowband interference in a time-varying OFDM system. Background Art

[0002] As a multi-carrier modulation technology, orthogonal frequency division multiplexing (OFDM) has advantages such as anti-multipath fading and is widely used in various wireless communication fields. In these fields, such as underwater acoustic communication, the Doppler frequency shift caused by the high-speed movement between the transceiver is an inevitable influence. At the same time, due to natural environment or military factors, impulse interference and narrowband interference often exist in these application fields, seriously affecting the communication quality and performance. Therefore, in a complex and changeable communication environment, it is particularly important to find a signal processing scheme when Doppler and combined interference coexist.

[0003] The clipping method, the limiting method, and the hybrid method combining the two are common non-linear impulse interference cancellation methods. The suppression of narrowband interference mostly reduces the power leakage caused by the frequency mismatch between the narrowband interference and the subcarriers through frequency estimation, and then zeros it through peak detection. However, these methods will cause signal distortion and affect the demodulation of data, so it is necessary to estimate the interference amplitude more accurately to improve the communication performance. At the same time, for the situation where the two types of interference and carrier frequency offset coexist, they all affect each other, and the traditional methods are not applicable to this communication scenario. Therefore, a new solution needs to be proposed. Summary of the Invention

[0004] The object of the present invention is to propose a method for jointly eliminating impulse interference and narrowband interference in a time-varying OFDM system. For OFDM communication in a time-varying channel, the inter-carrier interference caused by the carrier frequency offset not only deteriorates the demodulation performance of the useful data, but also affects the estimation and elimination of impulse interference and narrowband interference; at the same time, the existence of the two combined interferences also has a certain impact on the transmitted data and the estimation of the carrier frequency offset frequency. In order to improve the robustness of the system, a reasonable and effective way needs to be found to compensate for or eliminate the carrier frequency offset and combined interference to improve the communication quality. In view of this, the technical solution of the present invention is to provide a method for jointly eliminating impulse interference and narrowband interference in a time-varying OFDM system, which respectively completes carrier frequency offset compensation and combined interference elimination. Specifically, first, a Doppler-insensitive signal is used as a preamble for the estimation and compensation of the carrier frequency offset; then, the center frequency and the corresponding amplitude of the narrowband interference are estimated in the frequency domain, the narrowband interference is reconstructed and eliminated; finally, aiming at the sparsity of the impulse interference in the time domain, the impulse interference is estimated and eliminated through the measurement of the empty subcarriers.

[0005] The solution of the present invention is used for a time-varying OFDM system, and the processing of the signal transmitting end in the system includes:

[0006] The source bit data to be transmitted is subjected to symbol mapping. The constellation mapping symbols are subjected to serial-to-parallel conversion (S / P), OFDM modulation, and then parallel-to-serial conversion (P / S). The orthogonal phase shift keying modulation (QPSK) or orthogonal amplitude modulation (QAM) can be used for symbol mapping;

[0007] A preamble and zero padding (ZP) are added before and after each OFDM symbol respectively, as Figure 1 shown, and finally transmitted. The preamble is a Doppler-insensitive signal, and signals such as linear frequency modulation (LFM) signals or hyperbolic frequency modulation (HFM) signals can be used;

[0008] This invention mainly focuses on the processing of received signals. Therefore, the signal transmission process is omitted in the solution. The processing of received signals specifically includes:

[0009] S1. After using the preamble to complete carrier frequency offset estimation and compensation for the received signal, interference cancellation is performed. The object of cancellation is the new interference component in the compensated signal. It can be obtained theoretically that it is still the superposition of impulse interference and narrowband interference, specifically:

[0010] Using the LFM signal as the preamble and using the ambiguity function of the LFM signal for frequency offset estimation:

[0011]

[0012] where y L [n L is the received LFM signal, x L [n L is the transmitted LFM signal, f △ is the OFDM subcarrier spacing, N f is the quantization number, T LFM is the duration of the LFM signal, T s is the sampling time interval;

[0013] The estimated carrier frequency offset frequency is The corresponding carrier frequency offset matrix is The signal obtained after compensating the carrier frequency offset is expressed as:

[0014]

[0015] where y is the received signal after discrete sampling, H is the Toeplitz matrix, F pre is the precoding matrix, X is the transmitted frequency-domain OFDM signal, and They are impulse interference and narrowband interference respectively, and w is zero-mean additive white Gaussian noise;

[0016] S2. For the existing combined interference, first detect and estimate the peak frequency and corresponding amplitude of the signal in the frequency domain, reconstruct the narrowband interference and eliminate it. Specifically:

[0017] The narrowband interference is modeled as a single-tone interference:

[0018]

[0019] where P = N + N zp , N is the length of the OFDM symbol, N zp is the ZP length, A 0 is the interference amplitude, the center frequency of the interference is closest to the Mth subcarrier, and it follows a uniform distribution; Express in matrix form as:

[0020]

[0021] where only the Mth element is non-zero;

[0022] Estimate Λ and Z M to reconstruct Specifically:

[0023] Obtain the position corresponding to the maximum value point in the frequency domain according to peak detection, denoted as

[0024] Use the chirp-z transform to refine the spectrum of the single-tone interference in the frequency range of , let f △ be the subcarrier spacing, and define the frequency-domain signal after the chirp-z transform as Y CZT (l);

[0025] Take the l corresponding to the maximum value point of Y CZT (l) as l m , then the estimated value of m is:

[0026]

[0027] where N c is the number of points for spectrum refinement of the chirp-z transform, and obtain the corresponding estimate according to m

[0028] The frequency-domain expression of the signal after carrier frequency offset compensation is:

[0029]

[0030] Among them, F P is a P-dimensional Fourier transform matrix, is the external noise. Projected onto the left null space of V, that is such that UV = 0, we have:

[0031]

[0032] Among them Since Z M is a sparse vector, and the positions of the non-zero values are known, the least squares algorithm is used to estimate it, and Z M The estimated value of the non-zero value is

[0033]

[0034] where B is the column of UA; the finally estimated tone interference is The signal after tone interference cancellation is:

[0035]

[0036] S3. Use the Turbo compressive sensing algorithm to estimate the impulse interference, and subtract the impulse interference from the obtained signal to obtain the signal after joint interference cancellation, specifically:

[0037] Define that the impulse interference follows a Bernoulli-Gaussian distribution. For an impulse interference vector of length P each of its elements The probability density function is:

[0038]

[0039] Among them is the sparsity of the impulse interference, represents the complex Gaussian probability density function, whose mean is 0 and variance is N i ; define the set of subcarrier labels of the transmitted signal as Γ, the number of subcarriers is N null , define a selection matrix each column of which has only one element as 1 and the rest are 0, and the position of 1 depends on the subcarrier label; in the ZP-OFDM system, the method of tail data coverage and superposition is used to maintain the orthogonality between subcarriers. Let the respective signals be become become become and the sparsity becomes Then we have:

[0040]

[0041] where F N is an N-dimensional Fourier transform matrix, is zero-mean Gaussian white noise with a variance of N 0 , rewrite the above equation as:

[0042]

[0043] where Input into the Turbo compressive sensing module, and finally output the estimated impulse interference After impulse interference cancellation, the finally output signal is

[0044]

[0045] Then use the signal after impulse interference cancellation for signal detection.

[0046] The present invention aims at the carrier frequency offset and combined interference existing in a time-varying OFDM system, and proposes to first compensate the carrier frequency offset by using a preamble, then estimate and eliminate the narrowband interference according to the large amplitude characteristic of the narrowband interference in the frequency domain, and finally estimate and eliminate it by using the time-domain sparsity of the impulse interference through the measurement of null subcarriers.

[0047] The beneficial effect of the present invention is that the present invention is applicable to a mobile communication environment with combined interference, and the proposed method can achieve frequency offset compensation and interference cancellation, improving the communication quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is the structure of the transmitted OFDM symbol.

[0049] Figure 2 is the signal processing flow chart at the receiving end.

[0050] Figure 3 is the block diagram of the Turbo compressive sensing algorithm. DETAILED DESCRIPTION OF THE INVENTION

[0051] The following further describes in detail the specific embodiments of the present invention with reference to the drawings.

[0052] Define the amplitude and delay of the p-th path in the channel as A p and τ p , then the channel impulse response model is

[0053]

[0054] Among them, t c is the total number of multiple paths, T s is the sampling time interval, and δ[·] is the unit impulse function. The transmitted signal passes through the channel and is affected by carrier frequency offset, pulse and narrowband combined interference before reaching the receiving end. Define N as the length of the OFDM symbol, and N zp is the ZP length, P = N + N zp , then the received signal after discrete sampling is expressed as:

[0055] y = DHF pre X + i + z + w

[0056] Among them, is the received signal, is the transmitted OFDM signal in the frequency domain, is the pulse interference, is the narrowband interference, and w is the zero-mean additive white Gaussian noise with variance N 0 . The diagonal matrix D is the carrier frequency offset matrix, and there is where f d is the carrier frequency offset frequency, and f △ is the subcarrier spacing. is the precoding matrix, where, is the N-dimensional Fourier transform matrix, is an N×N zp dimensional all-zero matrix. is a Toeplitz matrix, and the elements of the first column are The elements of the first row are [h[0], 0 1×(P-1) .

[0057] (1) Carrier frequency offset compensation:

[0058] Use the LFM signal as the preamble. Define the bandwidth and duration of the LFM signal as B and T LFM , then the LFM signal can be expressed as

[0059]

[0060] Among them f 0 and K = B / T LFM are the starting frequency and the frequency sweep parameter of the LFM signal respectively. The ambiguity function of the LFM signal is

[0061]

[0062] Among them y L [n L is the received LFM signal, f △is the OFDM subcarrier spacing, and N f is the number of quantization points. Then the estimated carrier frequency offset is The corresponding carrier frequency offset matrix is where The signal obtained after compensating for the carrier frequency offset can be expressed as

[0063]

[0064] where and are still impulse interference and narrowband interference respectively. The subsequent steps will directly estimate and eliminate the changed impulse interference and narrowband interference for estimation and elimination.

[0065] (2) Narrowband interference cancellation:

[0066] The interference can usually be modeled as a single-tone interference, that is

[0067]

[0068] where A 0 is the interference amplitude, the center frequency of the interference is closest to the Mth subcarrier, and it follows a uniform distribution. For the interference it can also be expressed in matrix form as

[0069]

[0070] where F P is a P-dimensional Fourier transform matrix, and only the Mth element in it is non-zero.

[0071] It is necessary to estimate Λ and Z M to reconstruct For Λ, it is mainly divided into the following steps:

[0072] The first step: Obtain the position corresponding to the maximum value point in the frequency domain according to peak detection, denoted as

[0073] The second step: Since therefore, the chirp-z transform (CZT) can be used to refine the spectrum of the single-tone interference in the frequency range of Let The frequency-domain signal Y CZT after the CZT transform is as follows:

[0074]

[0075] where N c is the number of points for spectrum zooming, W 1 represents the radius of the starting sampling point of the zoomed spectrum, φ represents the phase corresponding to the starting frequency sampling point, and take W 1 = 1, φ = 2πf 1 / f s ; V 1 is the stretching rate of the sampling line, θ is the angular frequency interval between adjacent sampling points, and take V 1 = 1,

[0076] Step 3: Take the l corresponding to the maximum point of Y CZT (l) as l m , then the estimated value of m is

[0077]

[0078] According to the corresponding estimation can be obtained

[0079] Another quantity to be estimated for single-tone interference is Z M . The frequency-domain expression of the signal after carrier frequency offset compensation is

[0080]

[0081] where, is the external noise. To reduce the influence of the signal on the interference amplitude estimation, project onto the left null space of V, that is make UV = 0, then there is

[0082]

[0083] where Since Z M is a sparse vector, and the positions of the non-zero values are known, the least squares algorithm can be used to estimate it. The estimated value of the non-zero value of Z M is

[0084]

[0085] where B is the th column of UA. The finally estimated single-tone interference is The signal after single-tone interference cancellation is

[0086]

[0087] (3) Pulse interference cancellation:

[0088] The considered impulse interference follows a Bernoulli-Gaussian distribution, that is, for an impulse interference vector of length P The probability density function of each of its elements is

[0089]

[0090] where is the sparsity of the impulse interference, represents the complex Gaussian probability density function with a mean of 0 and a variance of N i . Suppose the set of subcarrier labels of the transmitted signal is Γ, and the number of empty subcarriers is N null . Define a selection matrix Each column of which has only one element equal to 1 and the rest equal to 0, and the position of 1 depends on the empty subcarrier label. To accurately extract the empty subcarrier positions, for the described ZP-OFDM system, the method of overlaying the tail data coverage can be used to maintain the orthogonality between subcarriers. At this time, each part of the signal is respectively changed from to changed to changed to and its sparsity becomes Then there is

[0091]

[0092] where is white Gaussian noise with zero mean and variance N 0 . The above equation can be rewritten as

[0093]

[0094] will enter the Turbo compressive sensing module as the input, and finally the estimated impulse interference is output The Turbo compressive sensing algorithm is as shown in Figure 3 and is roughly divided into two modules:

[0095] Module A: This module is an LMMSE estimator. In this module, each element in the signal follows an i.i.d Gaussian prior distribution, and its prior mean and variance are respectively and Then The prior mean of is

[0096]

[0097] The prior variance is Since S is a partial orthogonal matrix, the LMMSE estimate and its corresponding variance are respectively

[0098]

[0099]

[0100] and we have

[0101]

[0102] Then the extrinsic information output by module A about is

[0103]

[0104]

[0105] Module B: This module is an MMSE estimator, which mainly uses the sparse structure of the signal for further accurate estimation. Model as the additive Gaussian white noise observation value of the sparse signal i.e.,

[0106]

[0107] where and is independent of i. From module A, we can obtain

[0108]

[0109] Then the MMSE estimate and its corresponding variance are

[0110]

[0111]

[0112] where var[pq] = E[p - E[pq] 2 q]. Then we have

[0113]

[0114] Furthermore, we can obtain

[0115]

[0116]

[0117] For the next iteration, we can give

[0118]

[0119] After the iteration terminates, Will be used as The estimated result. After pulse interference cancellation, the finally output signal is

[0120]

[0121] Then signal detection is performed using the signal after pulse interference cancellation.

Claims

1. A method for jointly eliminating impulse interference and narrowband interference in a time-varying OFDM system, characterized in that, it includes the following steps: S1. First, use the preamble to complete carrier frequency offset estimation and compensation for the received signal, and then perform interference cancellation. The object of cancellation is the new interference component in the compensated signal. Specifically: Use the LFM signal as the preamble and use the ambiguity function of the LFM signal for frequency offset estimation: where y L [n] is the received LFM signal, and x L [n] is the transmitted LFM signal, f △ is the OFDM subcarrier spacing, N f is the quantization number of points, T LFM is the duration of the LFM signal, and T s is the sampling time interval; The estimated carrier frequency offset is The corresponding carrier frequency offset matrix is The signal obtained after compensating for the carrier frequency offset is expressed as: where y is the received signal after discrete sampling, H is the Toeplitz matrix, F pre is the precoding matrix, X is the transmitted frequency-domain OFDM signal, and are impulse interference and narrowband interference respectively, and w is zero-mean additive white Gaussian noise; S2. For the existing joint interference, first detect and estimate the peak frequency and corresponding amplitude of the signal in the frequency domain, reconstruct the narrowband interference and perform cancellation. Specifically: Model the narrowband interference as a single-tone interference: Wherein, P = N + N zp , N is the length of the OFDM symbol, and N zp is the length of the ZP, A 0 is the interference amplitude, and the center frequency of the interference is closest to the M-th subcarrier and obeys a uniform distribution; is expressed in the form of a matrix as: will Among them, only the M-th element is non-zero; For Λ and Z M perform an estimation to reconstruct Specifically: Obtain the position corresponding to the maximum value point in the frequency domain according to peak detection, denoted as Use the chirp-z transform to refine the spectrum of the single-tone interference in the frequency band with a range of . Let f △ be the subcarrier spacing, and define the frequency-domain signal after the chirp-z transform as Y CZT (l); Take Y CZT (l) The l corresponding to the maximum point is l m , then the estimated value of m is: where N c is the number of points for spectrum refinement of the chirp-z transform, and the corresponding estimate is obtained according to m The frequency-domain expression of the signal after carrier frequency offset compensation is: where is external noise, and F P is a P-dimensional Fourier transform matrix, projects onto the left null space of V, that is such that UV = 0, we have: Among them Since Z M is a sparse vector and the positions of the non-zero values are known, the least squares algorithm is used to estimate it, and the estimated value of the non-zero value of Z M is where B is the th column of UA; the finally estimated tone interference is The signal after tone interference cancellation is: S3. Estimate impulse interference using the Turbo compressive sensing algorithm, and subtract the impulse interference from the obtained signal to obtain the signal after joint interference cancellation, specifically: Define that the impulse interference follows a Bernoulli-Gaussian distribution. For an impulse interference vector of length P each of its elements has the following probability density function: Among them is the sparsity of the impulse interference, represents the complex Gaussian probability density function with a mean of 0 and a variance of N i ; define the set of subcarrier labels of the transmitted signal as Γ, and the number of empty subcarriers as N null , and define a selection matrix each column of which has only one element equal to 1 and the rest equal to 0, and the position of 1 depends on the empty subcarrier label; in the ZP-OFDM system, the method of covering and superimposing the tail data is adopted to maintain the orthogonality between subcarriers, and let each part of the signal be respectively changed from to to to and the sparsity becomes Then there is: where F N is an N-dimensional Fourier transform matrix, is zero-mean Gaussian white noise with variance N 0 , and rewrite the above equation as: Among them Input into the Turbo compressive sensing module as the input, and finally output the estimated impulse interference After impulse interference cancellation, the finally output signal is Then use the signal after impulse interference cancellation for signal detection.

Images