Method, system and device for estimating delay inequality among multiple antenna signals in distributed system and medium

Abstract

The invention relates to the field of communication signal processing, in particular to a method, a system and equipment for estimating delay inequality among multiple antenna signals in a distributed system and a medium. Sparse Fourier transform (SFT) is introduced on the basis of a generalized quadratic correlation algorithm (GSCC), signals with sparsity in a time-frequency domain can be processed, the complexity of calculating signal Fourier transform is effectively reduced, and the traditional time delay inequality estimation performance based on a cross-correlation function is remarkably improved in a low signal-to-noise ratio environment. The invention analyzes the defects of the traditional weighting function, provides an improved generalized weighting function, can comprehensively consider the influence of two paths of signals, effectively overcomes the multi-peak phenomenon caused by the broadening of the peak value of a correlation function, and is particularly suitable for processing low-power signals. Compared with a traditional method, the time delay estimation precision and the calculation efficiency are improved, meanwhile, higher adaptability and universality are achieved, and the method is suitable for various distributed signal processing tasks.

Description

Technical Field

[0001] The present invention belongs to the field of communication technologies, and particularly relates to a method, system, device and medium for estimating the time delay difference between multi-antenna signals in a distributed system. Background Art

[0002] With the rapid development of wireless communication technologies and the increasing demand for information sharing efficiency, wireless communication has been more and more widely popularized and applied, making the electromagnetic environment more complex and changeable. A key problem that needs to be solved urgently in modern communication is how to ensure the reliable reception of weak signals in a low signal-to-noise ratio environment. In most cases, to meet the requirements of simple, reliable and low-cost projects, scholars have proposed to deploy multiple receiving devices in space through a distributed cooperation method, and use the multi-antenna signal-level synthesis technology to achieve the same reception effect as a large-aperture antenna, improve the signal-to-noise ratio of the received signal, and thus achieve the reliable reception of weak signals.

[0003] However, since the receiving devices in the distributed system are independent of each other and randomly deployed, the amplitude-phase-frequency responses of each receiving antenna are different, and the channel characteristics of each receiver are also different. When the same target signal arrives at each receiving antenna, there will inevitably be obvious differences in time delay, center frequency, phase and other parameters. These parameter differences will seriously affect the coherence of the received signal and the final synthesis efficiency of the system. Therefore, before realizing the collaborative reception and synthesis processing of the distributed system, it is first necessary to solve the problems of estimating and compensating various parameter differences between signals in order to perform coherent addition on multi-antenna signals. Among them, how to estimate the time difference of arrival is the key and difficulty for finally effectively realizing the synthesis.

[0004] Regarding the estimation and compensation of the time delay difference between signals, the Mao Feilong team proposed a time delay estimation and compensation scheme for antenna array signals based on the GPU platform to meet the real-time synthesis requirements of broadband, high-speed, and parallel signals. After integer time delay compensation, the signals are split into subbands, and the fractional time delay between antennas is calculated and compensated by the time delay difference estimation module through iterative rounds (Mao Feilong, Jiao Yiwen, Ma Hong, et al. Time Delay Compensation Method for Antenna Array Signals Based on GPU [J]. Systems Engineering and Electronics, 2023, 45(08): 2383-2394.). Yanping Liao and Shuang Liu proposed an improved least mean square time delay difference estimation algorithm to solve the time delay difference estimation problem under the influence of correlated Gaussian noise. First, the correlated Gaussian noise is filtered by calculating the auto-fourth-order cumulant and cross-fourth-order cumulant of the signals, and then a variable step size estimation model is designed based on the ArcTanGent function to improve the time difference of arrival (TDOA) estimation accuracy in a low signal-to-noise ratio environment (Liao, Yanping, and Shuang Liu. An Improved Variable Step Size LMSTDE Algorithm Based on Fourth-order Cumulant [C]. 2021 7th International Conference on Computer and Communications (ICCC). IEEE, 2021.). SU Ting and GAO Yong proposed a blind separation-based technique to solve the time delay difference problem under co-frequency interference. The algorithm first performs blind separation on the mixed received signals through particle filtering to suppress the influence of co-frequency interference on the estimation accuracy, and then uses the cross-correlation method to estimate the TDOA (Ting, Su, and G.A.O. Yong. TDOA estimation of dual-satellites interference localization based on blind separation [J]. Journal of Systems Engineering and Electronics, 2019, 30(4): 696-702.). Xiang Bo et al. based on the variable step size LMS algorithm, gave relevant simulation experiments for different gradient descent methods to select the optimal algorithm, and achieved fractional time delay difference estimation through sinc function interpolation (Xiang Bo, Zhong Dianru, Zhang Xuejun, et al. Research on Adaptive Filter Time Difference Estimation Based on Least Mean Square [J]. Journal of Chengdu University (Natural Science Edition), 2020, 39(02): 170-175.).Zhang Yu and Yan Tianfeng addressed the problem of low accuracy in time-delay difference estimation in low signal-to-noise ratio environments. They introduced sparse Fourier transform into the generalized cross-correlation algorithm. By performing SFFT on the signal and then applying the generalized cross-correlation algorithm, they improved the anti-noise and parameter estimation capabilities of the algorithm while ensuring the operation speed and reducing the computational complexity (Zhang Yu, Yan Tianfeng. Time-delay Estimation Algorithm Based on Sparse Fourier Transform of Generalized Cross-Correlation [J]. Electronics Optics & Control, 2019(3):5.).

[0005] Through the above analysis, the problems and defects existing in the existing technologies are as follows:

[0006] (1) In the field of time-delay difference estimation between signals, although the current time-delay difference estimation method based on the traditional cross-correlation function performs well in high signal-to-noise ratio environments, its anti-noise performance is poor, and the data storage and transmission costs are high, and the computational efficiency is low.

[0007] (2) In the current algorithms for time-delay difference estimation between signals, the existing generalized weighting function will broaden the peak of the cross-correlation function, generating multiple peaks and causing incorrect estimation; secondly, large errors may be introduced when processing signals with low power levels. Summary of the Invention

[0008] In order to overcome the problems existing in the above-mentioned existing technologies, the purpose of the present invention is to provide a method, system, medium and device for estimating the time-delay difference between multi-antenna signals in a distributed system; taking the reception model of two signals as an example, sparse Fourier transform is introduced on the basis of the traditional generalized cross-correlation algorithm, and on this basis, the traditional weighting function is optimized, and the improved SFT-GSCC time-delay difference estimation method is used to estimate the time-delay difference between multi-antenna signals in a distributed system, improving the time-delay estimation accuracy of low-power signals.

[0009] In order to achieve the above purpose, the technical solution adopted by the present invention is: a method for estimating the time-delay difference between multi-antenna signals in a distributed system. Taking the reception model of two signals as an example, sparse Fourier transform is introduced on the basis of the traditional generalized cross-correlation algorithm; the traditional weighting function is optimized, and an improved SFT-GSCC time-delay difference estimation method is used to estimate the time-delay difference between multi-antenna signals in a distributed system.

[0010] The present invention provides a method for estimating the time-delay difference between multi-antenna signals in a distributed system, specifically including the following steps:

[0011] Step 1, taking the reception model of two signals as an example, sparse Fourier transform is introduced on the basis of the traditional generalized cross-correlation algorithm;

[0012] Step 2: On the basis of Step 1, optimize the traditional weighting function, and use the improved SFT-GSCC time delay difference estimation method to estimate the time delay difference between multi-antenna signals in a distributed system.

[0013] In Step 1, taking the reception model of two signals as an example, introduce the sparse Fourier transform on the basis of the traditional generalized cross-correlation algorithm. The specific process is as follows:

[0014] For the convenience of analysis, only consider the time delay estimation problem between two signals. Assume that the transmitted signal of the signal source is s(t), and the received signals at receiving device 1 and receiving device 2 after wireless transmission can be expressed as:

[0015]

[0016] where, A 1 and A 2 respectively represent the amplitude parameters of the transmitted signal, that is, the amplitude gains after transmission through different transmission channels; n 1 (t) and n 2 (t) are zero-mean additive white Gaussian noise; τ 1 and τ 2 respectively represent the signal time delays caused after passing through different channels, and there is τ 1 ≤τ 2 .

[0017] Taking x 1 (t) as a reference for normalization, we get

[0018]

[0019] where, t = 0, 1, …, N - 1; A = A 2 / A 1 represents the amplitude ratio of the two received signals; d = τ 2 -τ 1 represents the time delay of the same source signal arriving at the two receiving stations, that is, the time delay value to be obtained; N represents the number of sampling points.

[0020] The sparse Fourier transform is used to process signals with sparsity in the time-frequency domain. The implementation method is to perform the Fourier transform on the down-sampled signal to determine the positions and values of non-zero coefficients or important coefficients. Then, set the important points in the Fourier transform output sequence as valid points and the rest as zero values, and thus obtain the Fourier transform of the signal. The specific process is as follows:

[0021] S11. Spectrum rearrangement: To reduce the complexity of calculating the Fourier transform of a signal, the sparse Fourier transform introduces frequency-domain binning. Its core idea is to hash and place the Fourier coefficients of the signal in the frequency domain into a set of B bins according to a certain rule Γ(·), as shown below. The black squares in the figure represent non-zero Fourier coefficients, and the white squares represent zero-valued Fourier coefficients. To avoid the collision problem that occurs when two or more large-value points enter the bins due to the relatively concentrated input signal frequencies during actual processing, it is necessary to perform spectrum rearrangement on the input signal.

[0022] According to the displacement and scaling properties of the signal, it can be known that when the signal is rearranged in the time domain, the same rearrangement effect will also be achieved in the frequency domain. Therefore, the time-domain signal is rearranged.

[0023] The displacement property of the signal can be expressed as: If then Y(k) = X(k + b). Where b is an arbitrary integer; N represents its length, and n ∈ [0, N - 1].

[0024] The scaling property of the signal can be expressed as: If y(n) = x(σn), then Y(k) = X(σ -1 k). Where σ -1 represents the modular inverse element of σ with respect to N. If coordinate overflow occurs during the transformation, take the modulus with respect to N.

[0025] Based on the above properties, the definition of the time-domain rearrangement of a signal x(n) with length N is

[0026] (P σ,τ x) n = x σn+τ

[0027] Where, assuming that the parameter σ is the scaling factor and is invertible for the operation of N; the integer τ is the displacement factor, and τ ∈ [0, N - 1]. Then the corresponding frequency-domain signal can be expressed as

[0028]

[0029] S12. Window function filtering: In the sparse Fourier transform, to reduce the operation time of calculating the signal spectrum, it is necessary to truncate the original signal and obtain the signal spectrum by calculating some input signal points. Usually, a smoothing window function is used to process the signal. This window function has the characteristics of a flat passband and small ripple, and can ensure the accuracy of the signal when extracting the signal.

[0030] Let f(n) be a window function with length N, which has only a finite number of non-zero elements in the time domain, and F(k) represents its frequency domain. If it satisfies the following conditions:

[0031]

[0032] Then \(F\) is called a flat window function with parameters \((\varepsilon,\varepsilon',\delta,\omega)\).

[0033] Among them, \(\text{supp}(f)\) represents the support set of the vector \(f(n)\); \(K\) represents the sparsity of the signal; \(B\) represents the window length, and there is \(\delta\) is the ripple of the oscillation, representing the fluctuation characteristic of the stopband, and \(\delta\approx1 / N\); \(\varepsilon\) is the stopband truncation factor, representing the ratio of the stopband cut-off frequency to the sampling frequency, and \(\varepsilon = 1 / 2B\); \(\varepsilon'\) is the passband truncation factor and \(\varepsilon'=2\varepsilon\). It can be observed from the above formula that the proposed smoothing window function is actually a low-pass filter that specifies the passband and stopband characteristics.

[0034] In the specific implementation, the standard window function approaches a rectangular window in the frequency domain. According to the duality of the time domain and the frequency domain, the sinc function \(h(n)\) in the time domain corresponds to the rectangular window sequence in the frequency domain. Let:

[0035]

[0036] Among them, \(f\) c is the cut-off frequency. Since the passband width is required to be \(N / B\), then

[0037]

[0038] In the SFT method, the filter is used to extract a part of the Fourier coefficients. A filter that can concentrate energy in the time domain and the frequency domain, has a narrow passband and stopband, and is as flat as possible in the passband needs to be designed. Since the frequency domain convolution of the signal is equivalent to the time domain product, the flat window function can be obtained by using the product of the time domain Gaussian window function and the sinc function.

[0039] In actual operation, usually an \(M = O(B\log N)\) point truncation operation needs to be performed. First, set the first \(\omega / 2\) and the last \(\omega / 2\) points of the all-zero sequence with length \(N\) to 1, then perform the inverse Fourier transform and intercept the last \(M / 2\) and the first \(M / 2\) points, and set the remaining \(N - M\) points to zero to obtain the corresponding \(N\)-point long standard window function graph.

[0040] The Gaussian window function can be generated with the standard deviation There is

[0041]

[0042] From this, the first \(M\) points can be obtained. After padding \(N - M\) zeros, the \(N\)-point Gaussian window function is obtained. Let

[0043] \(f(n)=h(n)\cdot g(n)\)

[0044] Obtain the target flat window function, and the corresponding spectrum F(k) = H(k) * G(k). By introducing the Gaussian window, the discontinuity at the truncation point can be effectively reduced, and the passband in the frequency domain is nearly flat, approaching the ideal state of the filter.

[0045] S13. Frequency-domain downsampling: According to the properties of the Fourier transform, aliasing in the time domain of the signal is equivalent to downsampling in the frequency domain. Assume the length of the complex signal x(n) is N, and B is an integer that can divide N. The signal frequency domain is downsampled at equal intervals of N / B to obtain y(n).

[0046]

[0047] Its Fourier transform is

[0048] Y(k) = X(k·N / B)

[0049] where X(k) is the Fourier transform of the signal x(n).

[0050] S14. Reconstruction: In the overall SFT algorithm, the purpose of spectrum rearrangement, filtering, and downsampling FFT is to convert the N-point DFT into a B-point DFT, and then the complete spectrum of the signal is reconstructed from the B-point DFT through the reconstruction algorithm. Among them, the positioning process is used to determine the position of the non-zero coefficients in the spectrum diagram in the original Fourier spectrum. According to the positions of the non-zero coefficients in the downsampling FFT, the coordinates of the large-value points are inversely mapped to the original signal spectrum, and through L rounds of cyclic iteration, the top k positions with the highest frequencies are extracted as the final positioning result. Then, through the estimation cyclic process, the non-zero coefficients or important coefficient values are estimated. The frequency values of the final large-value points are obtained by taking the median of the real and imaginary parts of all the estimation cyclic results respectively, eliminating the influence of phase changes and filter ripple in the downsampling FFT calculation results, and improving the accuracy of the required signal spectrum.

[0051] In the second step, optimize the traditional weighting function, and use the improved SFT-GSCC time-delay difference estimation method to estimate the time-delay difference between multi-antenna signals in a distributed system. The specific process is as follows:

[0052] The time-delay estimation algorithm process of generalized quadratic correlation is as follows: First, perform autocorrelation operation on the received signal x 1 to obtain R 11 , and at the same time perform cross-correlation operation on two received signals to obtain R 12 . Then, perform correlation operation on the results of the autocorrelation and cross-correlation operations again to obtain the quadratic correlation function. Multiply the quadratic correlation function by the generalized weighting function to obtain the power spectrum function After performing inverse Fourier transform and then peak detection, the time-delay estimation value is obtained. Four weighting functions are mainly considered and briefly analyzed below:

[0053] ROTH weighting: Its principle is equivalent to Wiener filtering, which can effectively suppress the high-frequency noise band, but may broaden the peak of the cross-correlation function. Its expression is

[0054]

[0055] Smoothing cross-correlation transform (SCOT) weighting: Compared with ROTH weighting, the SCOT weighting function can take into account the influence of two channels simultaneously, but it will still broaden the peak of the cross-correlation function. Its expression is

[0056]

[0057] Phase transform (PHAT) weighting: It is equivalent to whitening filtering, which can effectively suppress noise interference and sharpen the main peak of the cross-correlation function. However, for signals with small energy, it may cause the denominator to approach 0, resulting in large errors. Its expression

[0058]

[0059] Analysis of some commonly used weighting functions in the generalized cross-correlation algorithm shows that: The ROTH weighting function can effectively suppress the high-frequency noise band, but may broaden the peak of the correlation function, generating multiple peaks and causing false estimation; The SCOT weighting function, although it can take into account the influence of two receiving channels, when the power spectral densities of the two received signals are equal, it will cause the broadening of the correlation function peak, generating false peaks and causing false estimation; The PHAT weighting function has a strong suppression effect on reverberation interference due to the fixed constant in the numerator of its weighting function, but may introduce large errors when processing signals with low power levels. To address the above problems, a new generalized weighting function is proposed, and its expression is:

[0060]

[0061] It can be seen from the expression that compared with the ROTH weighting function that only weights with the sub-power spectral density function of a single-channel signal, the proposed weighting function takes into account the influence of two-channel signals and weights with the power spectral density functions of two-channel signals; Compared with the SCOT weighting function, the proposed weighting function overcomes the phenomenon of multiple peaks caused by the broadening of the correlation function peak; Compared with the PHAT weighting function, the proposed weighting function adds the power spectral density functions of two-channel signals to the numerator, which can effectively suppress the large errors caused by the denominator approaching zero, and the numerator will gradually decrease as the signal power decreases, which can adapt to low-power signals and improve the time-delay estimation accuracy.

[0062] Based on the above principle analysis, a sparse Fourier transform time-delay estimation algorithm based on generalized cross-correlation is proposed. The algorithm steps are as follows:

[0063] S21. Perform time-domain aliasing on the received signals x 1 (n) and x 2 (n) to obtain signals x 1 '(n) and x' 2 (n), that is, add them after taking the modulo congruence of N / p in the signal time domain. Here, p is the downsampling factor and N is the number of sampling points, to obtain

[0064]

[0065] S22. Perform downsampling FFT on x 1 '(n) and x' 2 (n), then perform quadratic correlation processing, and perform generalized weighting on the cross-power spectrum to obtain the power spectrum function

[0066] S23. Perform inverse fast Fourier transform on to obtain the generalized quadratic cross-correlation time-domain function, perform peak detection on it to obtain p values corresponding to the spectral peaks, substitute these p values back into the original correlation function, and the abscissa corresponding to the maximum value obtained is the time-delay estimation value.

[0067] Another object of the present invention is to provide a system for estimating the time-delay difference between multi-antenna signals under the distributed system, which implements the method for estimating the time-delay difference between multi-antenna signals under the distributed system, including:

[0068] A sparse Fourier transform module, which is used to process signals with sparsity in the time-frequency domain in step one, so as to reduce the operation time for calculating the spectra of multi-antenna signals;

[0069] An improved generalized quadratic correlation time-delay difference estimation module, which is used to adopt an improved SFT-GSCC time-delay difference estimation method in step two to improve the estimation accuracy of the time-delay difference between low-power multi-antenna signals under the distributed system.

[0070] Another object of the present invention is to provide a computer device. The computer device includes a memory and a processor. When the computer program stored in the memory is executed by the processor, the processor executes the steps of the method for estimating the time-delay difference between multi-antenna signals under the distributed system.

[0071] Another object of the present invention is to provide a computer-readable storage medium, which stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the method for estimating the time-delay difference between multi-antenna signals under the distributed system.

[0072] Another object of the present invention is to provide an information data processing terminal, and the information data processing terminal includes the system for estimating the time-delay difference between multi-antenna signals under the distributed system.

[0073] Combined with the above technical solutions, the advantages and positive effects of the present invention are as follows:

[0074] First, based on the generalized cross-correlation algorithm, the present invention introduces the sparse Fourier transform to improve the estimation performance of the traditional time-delay difference estimation method based on the cross-correlation function in a low signal-to-noise ratio environment; analyzes the deficiencies of common weighting functions, and adopts an improved SFT-GSCC time-delay difference estimation method to estimate the time-delay difference between multi-antenna signals in a distributed system, improving the computational efficiency and estimation performance of the traditional cross-correlation algorithm; the invention fills the gap in this field.

[0075] The sparse Fourier transform used can be used to process signals with sparsity in the time-frequency domain, reducing the complexity of calculating the Fourier transform of the signal.

[0076] The proposed improved generalized weighting function can take into account the influence of two signals, overcome the multi-peak phenomenon caused by the broadening of the peak of the correlation function, and is suitable for low-power signals, which can improve the time-delay estimation accuracy.

[0077] The proposed improved SFT-GSCC time-delay difference estimation method can estimate the time-delay difference between multi-antenna signals in a distributed system with high accuracy.

[0078] The research of the present invention should focus on developing a time-delay difference estimation algorithm for multi-antenna signals with higher computational efficiency and better estimation performance in a low signal-to-noise ratio environment; the present invention explores the use of the sparse Fourier transform method and the generalized cross-correlation technology, and can estimate the time-delay difference between multi-antenna signals in a distributed system.

[0079] In summary, compared with the prior art, the research of the present invention explores the problem of using the sparse Fourier transform for generalized cross-correlation time-delay difference estimation, emphasizes the importance of solving the problem of using the sparse Fourier transform for the generalized cross-correlation algorithm in the time-delay difference estimation of multi-antenna signals in a distributed system; any task involving the time-delay difference estimation of multi-antenna signals in a distributed system can be estimated using the present invention; the use of the improved SFT-GSCC time-delay difference estimation method is more suitable for low-power signals than the traditional generalized cross-correlation algorithm, improving the time-delay estimation accuracy; the present invention has better performance and wider generality.

[0080] Second, aiming at the problems of high computational complexity, poor anti-noise ability and poor performance in a low signal-to-noise ratio environment existing in the time-delay difference estimation method of multi-antenna signals in a distributed system in the prior art, the present invention proposes a time-delay difference estimation method based on the sparse Fourier transform (SFT) and an improved weighting function. By introducing the sparsity characteristics of the signal and optimizing the generalized weighting function, the computational efficiency and anti-interference ability are significantly improved, and at the same time, the adaptability in the low-power signal scenario is enhanced.

[0081] The received signal is processed in the time - frequency domain by the sparse Fourier transform method, simplifying the high computational complexity of the traditional Fourier transform into the efficient calculation of sparse signals. By using techniques such as spectral rearrangement, window function filtering, and downsampled Fourier transform, redundant calculations are significantly reduced, and the computational efficiency in complex distributed systems is optimized.

[0082] The improved weighting function comprehensively considers the intensity characteristics and frequency distribution characteristics of multi - antenna signals, and can effectively reduce the interference of noise and multi - peak phenomena. Compared with the traditional ROTH, SCOT, and PHAT weighting functions, the generalized weighting function proposed in the present invention exhibits better robustness in low signal - to - noise ratio and multi - interference environments.

[0083] By combining the sparse Fourier transform and the optimized weighting function, the present invention can accurately estimate the signal time - delay difference in low - power signal scenarios. The sparsity feature of the sparse Fourier transform cooperates with the denominator constraint ability of the improved weighting function, avoiding the error accumulation in the processing of low - power signals by traditional methods.

[0084] The technical solution of the present invention is not only applicable to the time - delay difference estimation of multi - antenna signals, but also can be extended to other signal processing tasks in complex distributed systems. Its algorithm structure is simple and highly scalable, suitable for real - time applications in various engineering scenarios, providing important support for efficient and reliable signal processing in distributed systems.

[0085] Assume that the transmitted signal from the signal source reaches two receiving devices after wireless transmission, and is affected by different path gains, noise, and interference, resulting in a time - delay difference in the signal. The present invention provides a basis for subsequent calculations by normalizing the received signal and extracting the signal amplitude ratio and time - delay information.

[0086] Utilize the sparsity feature of the sparse Fourier transform to process signals with time - frequency sparsity. By spectral rearrangement, window function filtering, and downsampled Fourier transform, important coefficients in the Fourier spectrum are extracted and redundancies are removed, reducing the computational complexity of the Fourier transform.

[0087] Based on the deficiencies of traditional weighting functions, an improved generalized weighting function is proposed. This weighting function is weighted by jointly considering the power spectral density functions of multiple signals, suppressing the noise interference and multi - peak phenomena in traditional methods, and at the same time being suitable for the precise processing of low - power signals.

[0088] Combine the sparse Fourier transform with the improved generalized quadratic correlation algorithm, process the cross - power spectrum by inverse Fourier transform, and obtain the estimated value of the signal time - delay difference by using the peak detection method. This method can still maintain high precision and high robustness under low signal - to - noise ratio and multi - interference conditions. Description of the Drawings

[0089] Figure 1 It is a flow chart of a method, system, medium and device for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention.

[0090] Figure 2 It is a schematic structural diagram of a system for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention.

[0091] Figure 3 It is a flow block diagram of a method for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention.

[0092] Figure 4 It is a schematic diagram of the simulation experiment results of the improved generalized weighted function cross-correlation peak when the signal-to-noise ratio of the system for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention is SNR = 0 dB.

[0093] Figure 5 It is a schematic diagram of the simulation experiment results of the improved generalized weighted function cross-correlation peak when the signal-to-noise ratio of the system for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention is SNR = -5 dB.

[0094] Figure 6 It is a schematic diagram of the simulation experiment results of the improved generalized weighted function cross-correlation peak when the signal-to-noise ratio of the system for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention is SNR = -10 dB.

[0095] Figure 7 It is a schematic diagram of the simulation experiment results of the performance comparison of the time delay difference estimation of different algorithms of the system for estimating the time delay difference between multi-antenna signals in a distributed system provided by an embodiment of the present invention. Detailed implementation manners

[0096] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0097] Embodiment 1: Time delay difference estimation in a multi-antenna radar system

[0098] In a distributed multi-antenna radar system, multiple receiving antennas are installed to receive signals reflected by a target. The estimation of the target distance depends on the time delay difference between the received signals. The method of the present invention is implemented through the following steps:

[0099] 1. Signal reception and preprocessing:

[0100] Multiple receiving antennas in the system collect the signals reflected by the target, and the signals contain noise and multipath interference. The signals are digitally sampled and normalized, and two signals x 1 (t) and x 2 (t) are extracted.

[0101] 2. Sparse Fourier transform:

[0102] Perform sparse Fourier transform on x 1 (t) and x 2 (t), including spectrum rearrangement, window function filtering and downsampled Fourier transform, to extract the sparse features of the signals and reduce the computational complexity.

[0103] 3. Generalized weighted function optimization:

[0104] Use the improved weighted function to weight the cross-power spectrum of the two signals, and overcome the interference of noise and multi-peak phenomena on the time-delay estimation in the traditional method.

[0105] 4. Time-delay difference calculation:

[0106] Perform inverse Fourier transform on the weighted power spectrum function, and combine the quadratic correlation algorithm to detect the peak position of the power spectrum function to obtain the time-delay difference between the two signals.

[0107] Through this method, the radar system realizes high-precision target distance measurement in the low signal-to-noise ratio and complex interference environment, and the error is less than 10 nanoseconds.

[0108] Example 2: Time-delay difference estimation of synchronization signals in a wireless communication network

[0109] In a distributed wireless communication network, multiple base stations cooperate to provide seamless coverage. When synchronizing between base stations, it is necessary to accurately estimate the time-delay difference between synchronization signals. The method of the present invention can be used to solve this problem, specifically as follows:

[0110] 1. Synchronization signal reception:

[0111] Two base stations respectively receive the synchronization signals s 1 (t) and s 2 (t). The signals are affected by different path losses and noise interference during transmission, and the base stations digitize the signals and transmit them to the central control unit.

[0112] 2. Sparse Fourier transform processing:

[0113] Perform sparse Fourier transform on s 1 (t) and s 2 (t), avoid the collision of high-value points in the frequency domain through spectrum rearrangement, and use the Gaussian window function to reduce the influence of spectrum oscillation, so as to extract the key frequency components.

[0114] 3. Weighting Process and Delay Calculation:

[0115] The cross-power spectrum of the synchronization signal is weighted using an improved generalized weighting function to enhance the peak signal and reduce the influence of noise. Combining with the peak detection of the power spectrum function by the quadratic correlation algorithm, the delay difference is obtained.

[0116] 4. Synchronization Time Adjustment:

[0117] According to the delay difference value, the central control unit adjusts the time synchronization signal between base stations to ensure that all base stations maintain nanosecond-level synchronization.

[0118] This embodiment significantly improves the synchronization accuracy of the distributed wireless communication network, supporting more efficient cooperative communication and data transmission.

[0119] Aiming at the problems existing in the prior art, the present invention provides a method, system, medium and device for estimating the delay difference between multi-antenna signals under a distributed system. The present invention will be described in detail below with reference to the accompanying drawings.

[0120] This method first establishes a reception model for two paths of signals, taking the multi-antenna signals under a distributed system as the research object. At the receiving end, after the signals are collected by the antennas, they are affected by signal propagation paths, noise and other interference factors, resulting in a delay difference. Although the traditional generalized quadratic correlation algorithm (GSCC) can estimate the delay difference, its performance is limited in complex environments, especially in the case of low signal-to-noise ratio and low-power signals. Therefore, this method is based on improving the traditional algorithm and introduces the sparse Fourier transform (SFT) to improve the robustness and computational efficiency of the algorithm.

[0121] The sparse Fourier transform is an algorithm that can efficiently process sparse signals. By utilizing the sparsity of the signals, the present invention significantly reduces the computational complexity on the basis of the traditional Fourier transform by reducing redundant operations. The sparse Fourier transform can efficiently decompose the signals in the time-frequency domain, extract the main features, and thus calculate the correlation between the two paths of signals more accurately, providing a reliable basis for subsequent delay difference estimation.

[0122] During the delay difference estimation process, the traditional weighting function is easily affected by noise interference, resulting in the broadening of the peak of the correlation function, thus generating a multi-peak phenomenon and reducing the estimation accuracy. By analyzing the limitations of the traditional weighting function, the present invention proposes an improved generalized weighting function. The improved weighting function can comprehensively consider the intensity and frequency characteristics of the two paths of signals, enhance the identification ability of the target peak, and at the same time reduce the influence of noise on the result.

[0123] Based on the sparse Fourier transform and the improved weighting function, the present invention proposes an improved SFT-GSCC time delay difference estimation method. By combining the time-domain and frequency-domain information of the signal, this method optimizes the correlation calculation process of the traditional GSCC algorithm, enabling it to maintain a high time delay difference estimation accuracy even in a low signal-to-noise ratio environment. Meanwhile, the efficient calculation characteristics of the sparse Fourier transform further reduce the overall complexity of the algorithm.

[0124] The improved SFT-GSCC method is particularly suitable for processing low-power signals. Through the sparsity characteristics of the sparse Fourier transform and the denoising ability of the improved weighting function, the present invention can overcome the limitations of traditional methods in the low-power signal scenario, effectively improve the robustness and adaptability of time delay difference estimation, and make it widely applicable to various complex distributed systems.

[0125] This method is not only applicable to the time delay difference estimation between multi-antenna signals, but also can be extended to signal processing tasks in other distributed systems. Compared with traditional methods, the present invention shows significant advantages in terms of computational efficiency, estimation accuracy, and anti-noise ability. At the same time, due to its simple algorithm structure and strong scalability, it provides important support for practical engineering applications and has higher generality and practical application value.

[0126] Those of ordinary skill in the art in the industry can also implement the method for estimating the time delay difference between multi-antenna signals in a distributed system provided by the present invention by using other steps. Figure 1 The method, system, medium, and device for estimating the time delay difference between multi-antenna signals in a distributed system provided by the present invention are merely a specific embodiment.

[0127] A method for estimating the time delay difference between multi-antenna signals in a distributed system includes:

[0128] 1) Establish a reception model for multi-antenna signals;

[0129] 2) Use the sparse Fourier transform to perform time-frequency domain decomposition on the received signal and extract signal features;

[0130] 3) Construct an improved weighting function based on the extracted signal features;

[0131] 4) Apply the improved weighting function to the generalized cross-correlation algorithm and calculate the time delay difference by combining the time-domain and frequency-domain information of the signal.

[0132] The sparse Fourier transform efficiently processes the received signal by reducing redundant operations and extracts the main frequency components of the signal using sparsity characteristics.

[0133] The improved weighting function is optimized according to the intensity characteristics and frequency distribution characteristics of the signal, and can reduce the multi-peak phenomenon in the correlation function.

[0134] The improved SFT-GSCC time delay difference estimation method combines the efficient decomposition ability of the sparse Fourier transform and the optimization result of the improved weighting function to estimate the time delay difference of the received signal under low signal-to-noise ratio conditions.

[0135] It further includes the following steps:

[0136] 1) Perform sparse Fourier transform on the received signal to reduce the computational complexity by using its sparsity;

[0137] 2) Construct and optimize the generalized weighting function to enhance the target peak identification ability according to the signal strength and frequency characteristics;

[0138] 3) Use the improved sparse Fourier transform generalized cross-correlation (SFT-GSCC) algorithm to estimate the time delay difference between multi-antenna signals;

[0139] 4) Determine the time delay difference value through peak detection.

[0140] 6. The method according to claim 1, wherein the sparse Fourier transform includes the following steps:

[0141] 1) Perform spectral rearrangement on the signal to avoid the collision of high-value points in the frequency domain;

[0142] 2) Apply a smoothing window function to truncate the signal to reduce the frequency domain oscillation ripple and improve the passband flatness;

[0143] 3) Perform downsampling Fourier transform on the truncated signal to generate the downsampled spectrum.

[0144] The optimized generalized weighting function calculates the weights by jointly using the power spectral density functions of multiple paths of signals, reduces noise interference and multi-peak phenomena, and adapts to low-power signals to improve the time delay difference estimation accuracy.

[0145] The SFT-GSCC algorithm is implemented through the following steps:

[0146] 1) Perform downsampling Fourier transform on two paths of received signals and calculate the cross-power spectrum;

[0147] 2) Perform generalized weighting processing on the cross-power spectrum to obtain the power spectral function;

[0148] 3) Perform inverse Fourier transform on the power spectral function and perform peak detection to calculate the target time delay difference value.

[0149] As Figure 1 shown, the method for estimating the time delay difference between multi-antenna signals in the distributed system provided by the embodiment of the present invention is specifically as follows:

[0150] Step 1: Taking the reception model of two signals as an example, the sparse Fourier transform is introduced on the basis of the traditional generalized quadratic correlation algorithm. The specific process is as follows:

[0151] For the convenience of analysis, only the time delay estimation problem between two signals is considered. Suppose the transmitted signal of the signal source is s(t). After wireless transmission, the received signals at receiving device 1 and receiving device 2 can be expressed as:

[0152]

[0153] where, A 1 and A 2 respectively represent the amplitude parameters of the transmitted signal, that is, the amplitude gains after transmission through different transmission channels; n 1 (t) and n 2 (t) are zero-mean additive Gaussian white noises; τ 1 and τ 2 respectively represent the signal time delays caused after passing through different channels, and τ 1 ≤τ 2 .

[0154] Normalize with x 1 (t) as the reference to obtain

[0155]

[0156] where, t = 0, 1, …, N - 1; A = A 2 / A 1 represents the amplitude ratio of the two received signals; d = τ 2 -τ 1 represents the time delay of the same source signal arriving at the two receiving stations, that is, the time delay value to be obtained; N represents the number of sampling points.

[0157] The sparse Fourier transform is used to process signals with sparsity in the time-frequency domain. The implementation method is to perform Fourier transform on the downsampled signal to determine the positions and values of non-zero coefficients or important coefficients. Then, set the important points in the Fourier transform output sequence as the points with values and the rest as zero values, so as to obtain the Fourier transform of the signal. The specific process is as follows:

[0158] S11. Spectrum rearrangement: To reduce the complexity of calculating the Fourier transform of the signal, the sparse Fourier transform introduces frequency domain binning. Its core idea is to hash and place the Fourier coefficients in the frequency domain of the signal into a group of B bins according to a certain rule Γ(·), as shown below. The black squares in the figure represent non-zero Fourier coefficients, and the white squares represent zero Fourier coefficients. In order to avoid the collision problem that two or more large value points enter the bin due to the relatively concentrated input signal frequency in the actual processing process, it is necessary to perform spectrum rearrangement on the input signal.

[0159] According to the shift property and scaling property of signals, it can be known that when the signal is rearranged in the time domain, the same rearrangement effect will also be achieved in the frequency domain. Therefore, the time-domain signal is rearranged.

[0160] The shift property of the signal can be expressed as: If then Y(k) = X(k + b). Where b is an arbitrary integer; N represents its length, and n ∈ [0, N - 1].

[0161] The scaling property of the signal can be expressed as: If y(n) = x(σn), then Y(k) = X(σ -1 k). Where σ -1 represents the modular inverse element of σ with respect to N. If coordinate overflow occurs during the transformation process, take the modulus with respect to N.

[0162] Based on the above properties, the definition of time-domain rearrangement of the signal x(n) with length N is

[0163] (P σ,τ x) n = x σn+τ

[0164] Where it is assumed that the parameter σ is the scaling factor and is invertible for the operation of N; the integer τ is the shift factor, and τ ∈ [0, N - 1]. Then the corresponding frequency-domain signal can be expressed as

[0165]

[0166] S12. Window function filtering: In sparse Fourier transform, in order to reduce the operation time of calculating the signal spectrum, it is necessary to truncate the original signal and obtain the signal spectrum by calculating some input signal points. Usually, a smoothing window function needs to be used to process the signal. This window function has the characteristics of a flat passband and small ripple, and can ensure the accuracy of the signal when extracting the signal.

[0167] Let f(n) be a window function with length N, which has only a finite number of non-zero elements in the time domain, and F(k) represents its frequency domain. If it satisfies the following conditions:

[0168]

[0169] Then F is called a flat window function with parameters (ε, ε', δ, ω).

[0170] Where supp(f) represents the support set of the vector f(n); K represents the sparsity of the signal; B represents the window length, and there is δ is the oscillation ripple, representing the stopband fluctuation characteristic, and δ≈1 / N; ε is the stopband truncation factor, representing the ratio of the stopband cut-off frequency to the sampling frequency, and ε=1 / 2B; ε' is the passband truncation factor and ε'=2ε. It can be observed from the above formula that the proposed smoothing window function is actually a low-pass filter that specifies the passband and stopband characteristics.

[0171] In terms of specific implementation, the standard window function approaches a rectangular window in the frequency domain. According to the duality of time domain and frequency domain, the sinc function h(n) in the time domain corresponds to the rectangular window sequence in the frequency domain. Let:

[0172]

[0173] where f c is the cut-off frequency. Since the passband width is required to be N / B, then

[0174]

[0175] In the SFT method, the filter is used to extract a part of the Fourier coefficients. A filter that can concentrate energy in the time domain and frequency domain, has a narrow passband and stopband, and is as flat as possible in the passband needs to be designed. Because the frequency domain convolution of the signal is equivalent to the time domain product, the flat window function can be obtained by using the product of the time domain Gaussian window function and the sinc function.

[0176] In actual operation, an M=O(BlogN) point truncation operation is usually required. First, set the first ω / 2 and the last ω / 2 points of the all-zero sequence with length N to 1, then perform the inverse Fourier transform and intercept the last M / 2 and the first M / 2 points, and set the remaining N-M points to zero to obtain the corresponding N-point long standard window function graph.

[0177] The Gaussian window function can be generated with the standard deviation and there is

[0178]

[0179] From this, the first M points can be obtained. After padding N-M zeros, the N-point Gaussian window function is obtained. Let

[0180] f(n)=h(n)·g(n)

[0181] to obtain the target flat window function, and the corresponding spectrum F(k)=H(k)*G(k). By introducing the Gaussian window, the discontinuity at the truncation point can be effectively reduced, and the passband in the frequency domain is almost flat, approaching the ideal state of the filter.

[0182] S13. Frequency-domain downsampling: According to the properties of the Fourier transform, aliasing in the time domain of a signal is equivalent to downsampling in the frequency domain. Let the length of the complex signal x(n) be N, and B be an integer that can divide N. The signal frequency domain is downsampled at equal intervals of N / B to obtain y(n).

[0183]

[0184] Its Fourier transform is

[0185] Y(k) = X(k·N / B)

[0186] where X(k) is the Fourier transform of the signal x(n).

[0187] S14. Reconstruction: In the overall SFT algorithm, the purpose of spectrum rearrangement, filtering, and downsampling FFT is to convert the N-point DFT into a B-point DFT, and then the complete spectrum of the signal is reconstructed from the B-point DFT through a reconstruction algorithm. Among them, the positioning process is used to determine the position of the non-zero coefficients in the spectrogram in the original Fourier spectrum. According to the positions of the non-zero coefficients in the downsampling FFT, the coordinates of the large-value points are inversely mapped to the original signal spectrum, and through L rounds of iterative loops, the top k positions with the highest frequencies are extracted as the final positioning results. Then, through the estimation loop process, the non-zero coefficients or important coefficient values are estimated. By taking the median of the real and imaginary parts of all the estimation loop results respectively, the frequency values of the final large-value points are obtained, eliminating the influence of phase changes and filter ripple in the downsampling FFT calculation results, and improving the accuracy of the required signal spectrum.

[0188] Step 2: Optimize the traditional weighting function, and use the improved SFT-GSCC time-delay difference estimation method to estimate the time-delay difference between multi-antenna signals in a distributed system. The specific process is as follows:

[0189] The time-delay estimation algorithm process of the generalized cross-correlation is as follows: First, perform autocorrelation operation on the received signal x 1 to obtain R 11 , and at the same time perform cross-correlation operation on two received signals to obtain R 12 . Then, perform correlation operation on the results of the autocorrelation and cross-correlation operations again to obtain the quadratic correlation function. Multiply the quadratic correlation function by the generalized weighting function to obtain the power spectral function After inverse Fourier transform, peak detection is performed to obtain the time-delay estimation value. The following four weighting functions are mainly considered and briefly analyzed below:

[0190] ROTH weighting: Its principle is equivalent to Winener filtering, which can effectively suppress the high-frequency noise band, but may broaden the peak of the cross-correlation function. Its expression is

[0191]

[0192] Smoothing Correlation Transform (SCOT) weighting: The SCOT weighting function can take into account the effects of both channels compared to ROTH weighting, but it still broadens the peak of the cross-correlation function. Its expression is

[0193]

[0194] Phase Transform (PHAT) weighting: It is equivalent to whitening filtering, which can effectively suppress noise interference and sharpen the main peak of the cross-correlation function. However, for signals with relatively low energy, it may cause the denominator to approach 0, resulting in large errors. Its expression

[0195]

[0196] Analysis of some commonly used weighting functions in the generalized cross-correlation algorithm shows that: The ROTH weighting function can effectively suppress the high-frequency noise band, but it may broaden the peak of the correlation function, resulting in multiple peaks and causing false estimation; The SCOT weighting function can take into account the effects of both receiving channels, but when the power spectral densities of the two received signals are equal, it will lead to the broadening of the correlation function peak, generating false peaks and causing false estimation; Since the numerator of the PHAT weighting function is a fixed constant, it has a strong suppression effect on reverberation interference, but it may introduce large errors when processing signals with low power levels. To address the above problems, a new generalized weighting function is proposed, and its expression is:

[0197]

[0198] It can be seen from the expression that compared with the ROTH weighting function that only weights using the sub-power spectral density function of a single-channel signal, the proposed weighting function takes into account the effects of both signals and weights using the power spectral density functions of both signals; compared with the SCOT weighting function, the proposed weighting function overcomes the phenomenon of multiple peaks caused by the broadening of the correlation function peak; compared with the PHAT weighting function, the proposed weighting function adds the power spectral density functions of both signals to the numerator, which can effectively suppress the large errors caused by the denominator approaching zero, and the numerator will gradually decrease as the signal power decreases, which can adapt to low-power signals and improve the time-delay estimation accuracy.

[0199] Based on the above principle analysis, a sparse Fourier transform time-delay estimation algorithm based on generalized cross-correlation is proposed. The algorithm steps are as follows:

[0200] S21. Perform time-domain aliasing on the received signals x 1 (n) and x 2 (n) to obtain the signals x 1 '(n) and x' 2(n), that is, adding after taking the modulo congruence of N / p in the signal time domain, where p is the downsampling factor and N is the number of sampling points, to obtain

[0201]

[0202] S22. For x 1 '(n) and x' 2 (n), perform downsampling FFT and then perform quadratic correlation processing, and perform generalized weighting on the cross-power spectrum to obtain the power spectrum function

[0203] S23. For Perform inverse fast Fourier transform to obtain the generalized quadratic cross-correlation time-domain function, perform peak detection on it to obtain p values corresponding to the spectral peaks, substitute these p values back into the original correlation function, and the abscissa corresponding to the maximum value obtained is the time-delay estimation value.

[0204] As Figure 2 shown, the system of the method for estimating the time-delay difference between multi-antenna signals under a distributed system provided by the embodiment of the present invention includes:

[0205] Sparse Fourier transform module 1, which is used to process signals with sparsity in the time-frequency domain in step one, thereby reducing the operation time for calculating the spectrum of multi-antenna signals.

[0206] Improved generalized quadratic correlation time-delay difference estimation module 2, which is used to adopt an improved SFT-GSCC time-delay difference estimation method in step two to improve the estimation accuracy of the time-delay difference between low-power multi-antenna signals under a distributed system.

[0207] The time-delay difference method provided by the present invention can be used to estimate the time-delay difference between low-power multi-antenna signals under a distributed system.

[0208] The technical effects of the present invention will be described in detail below in combination with simulation experiments.

[0209] In order to evaluate the performance of the present invention, simulation verification is carried out. In the simulation experiment, in order to analyze the cross-correlation time-delay estimation waveform performance of the SCOT weighting function, PHAT weighting function, ROTH weighting function, and the proposed weighting function, two-channel received signals x 1 and x 2 added with stationary Gaussian white noise are constructed. The received signals used in the simulation are all QPSK signals, the sampling frequency is 10 kHz, the number of sampling points is 8192, and the relative time-delay value of x 1 and x 2 is set to 0.2 s, and the signal-to-noise ratios are taken as 0 dB, -5 dB, and -10 dB respectively.

[0210] From Figure 4It can be seen that when the signal-to-noise ratio SNR = 0 dB, the cross-correlation functions of the three traditional weighting functions all have relatively sharp peaks, and the estimation effects are not much different. However, the amplitude of the cross-correlation function of the improved generalized weighting function proposed in the present invention is higher than that of the other three methods. The delay estimation values of the four weighting functions are all τ = 0.2 s, which is consistent with the relative delay value of x 1 and x 2 set in the experiment, indicating that all four methods can obtain relatively high delay estimation accuracy. As can be seen from Figure 5 , when the signal-to-noise ratio SNR = -5 dB, the sharpening degree of the peaks of the other three weighting functions is smaller than that of the proposed weighting function, indicating that the improved generalized weighting function has certain advantages at low signal-to-noise ratios. As can be seen from Figure 6 , as the signal-to-noise ratio decreases, the delay estimation peaks of the SCOT weighting function, PHAT weighting function, and ROTH weighting function are more affected by noise and change greatly. At the signal-to-noise ratio SNR = -5 dB, their delay estimation peaks are almost submerged in the noise and are affected by secondary peaks, resulting in ineffective delay estimation. However, the proposed weighting function still has a relatively high peak and less influence from secondary peaks, and the delay estimation value τ = 0.2 s, with good delay estimation accuracy and anti-noise performance. The above simulation results show that the proposed weighting function can effectively suppress the influence of secondary peaks of the cross-correlation function, has a sharp peak, can effectively estimate the delay difference in a low signal-to-noise ratio environment, and has a relatively high delay estimation accuracy rate.

[0211] To further verify the delay estimation performance of the improved algorithm, MATLAB was used to simulate and compare the generalized cross-correlation, quadratic correlation, generalized quadratic correlation, and SFT-GSCC improvement algorithms in different signal-to-noise ratio environments. The simulation constructed two-channel QPSK received signals with added stationary Gaussian white noise, a sampling frequency of 100 kHz, 8192 sampling points, and a decimation factor p = 8. Assuming a delay of 200 sampling intervals between the two signals, that is, a delay time of 2 ms, Gaussian white noise with a signal-to-noise ratio range of -15 dB to 5 dB was added to it, with a noise interval of 1 dB. The root mean square error was used as the performance index, and its expression is as follows

[0212]

[0213] where τ 0 is the true delay estimation value, τ i is the i-th delay estimation value, and N is the total number of delay estimations. 1000 Monte Carlo simulations were performed on the four algorithms respectively. As can be seen from Figure 7It can be seen that the delay estimation performance of the four algorithms will all decline as the signal-to-noise ratio decreases. In a high signal-to-noise ratio environment, all four algorithms can effectively estimate the delay difference, and the estimation effects of the generalized quadratic correlation algorithm and the algorithm proposed in the present invention are close. When the SNR is in the range of [-5dB, 5dB], the RMSE of the generalized cross-correlation algorithm and the quadratic correlation algorithm will increase as the SNR decreases, while the changes of the generalized quadratic correlation and the improved SFT-GSCC algorithm are relatively slow; when the SNR is in the range of [-15dB, -5dB], the RMSE of the generalized cross-correlation algorithm, the quadratic correlation algorithm, and the generalized quadratic correlation algorithm will increase rapidly as the SNR decreases, while the change of the improved method is relatively slow, indicating that it can also have good delay estimation performance in a low signal-to-noise ratio environment. The above results show that regardless of whether the signal propagation channel environment is harsh or not, the improved SFT-GSCC delay estimation method proposed in the present invention can have good estimation accuracy and anti-noise performance. The simulation experiment verifies the creativity of the estimation method proposed in the present invention.

[0214] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those of ordinary skill in the art can understand that the above devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits of programmable hardware devices such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips and transistors, or programmable logic devices such as field programmable gate arrays, and can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above hardware circuits and software, such as firmware.

[0215] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention shall be covered by the protection scope of the present invention.

Claims

1. A method for estimating time delay difference between multi-antenna signals in a distributed system, characterized in that: The following steps are involved: - Perform time domain aliasing on the received signal and process the sparsity of the signal through sparse Fourier transform; - Optimize the traditional weighting function and use an improved sparse Fourier transform combined with generalized quadratic correlation to estimate the delay difference, including spectrum rearrangement, window function filtering, frequency domain downsampling and reconstruction.

2. The method for estimating the time delay difference between multiple antenna signals in a distributed system according to claim 1, characterized in that: The sparse Fourier transform determines the position and value of the non-zero coefficients in the frequency domain of the signal by performing Fourier transform on the downsampled signal; then the spectrum rearrangement operation is used to rearrange the signal in the time domain, and the important frequency information in the signal is retained through window function filtering, thereby achieving frequency domain downsampling.

3. The method for estimating the time delay difference between multiple antenna signals in a distributed system according to claim 1, characterized in that: Spectral rearrangement rearranges the Fourier coefficients of the signal in the frequency domain by redistributing the order of signal points in the time domain to avoid signal conflicts caused by high-frequency noise or energy concentration.

4. The method for estimating the time delay difference between multiple antenna signals in a distributed system according to claim 1, characterized in that: The window function filter uses a flat window function, whose time domain signal is the product of the Gaussian window function and the sinc function. The truncation operation is used to reduce the complexity of spectrum calculation and improve the flatness of the spectrum in the passband.

5. The method for estimating the time delay difference between multiple antenna signals in a distributed system according to claim 1, characterized in that: The quadratic correlation function is processed by a generalized weighting function, which is constructed based on a signal power spectrum density function, and can avoid high-frequency noise interference and the problem of the denominator approaching zero when the signal energy is low.

6. The method for estimating the time delay difference between multiple antenna signals in a distributed system according to claim 1, characterized in that: The downsampled frequency domain signal is subjected to inverse Fourier transform to extract the position corresponding to the spectrum peak. The final position of the non-zero coefficient is determined through multiple rounds of iterations, and the value of the non-zero coefficient is estimated to reconstruct the complete spectrum of the signal.

7. A system for estimating the time delay difference between multiple antenna signals in a distributed system, which implements the method for estimating the time delay difference between multiple antenna signals in a distributed system as claimed in any one of claims 1 to 6, characterized in that: include: The sparse Fourier transform module is used to perform spectrum rearrangement, window function filtering and downsampling Fourier transform on the received signal to extract the sparse features of the signal to reduce the computational complexity; The improved generalized quadratic correlation time delay difference estimation module is used to adopt an improved SFT-GSCC time delay difference estimation method in step 2 to improve the estimation accuracy of the time delay difference between low-power multi-antenna signals in a distributed system.

8. A device for estimating time delay difference between multiple antenna signals in a distributed system, characterized in that: include: a) Memory, used to store computer programs and data related to signal processing; b) a processor for executing a computer program in the memory, wherein the program comprises the following steps: i. Performing a sparse Fourier transform on the received signal to extract the sparse features of the signal; ii. Construct an improved generalized weighting function to weight the cross power spectrum; iii. Based on the improved sparse Fourier transform generalized quadratic correlation algorithm, the weighted power spectrum function is inversely Fourier transformed, and the delay difference estimation value of the signal is obtained through peak detection.

9. A computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to perform the steps of the method for estimating time delay differences between multiple antenna signals in a distributed system as described in any one of claims 1 to 6.

10. An information data processing terminal, comprising the system for estimating time delay differences between multiple antenna signals in a distributed system according to claim 7.

Images