Clutter suppression and sensing parameter estimation method in sensing mobile network

Abstract

The invention discloses a clutter suppression and sensing parameter estimation method in a sensing mobile network. The method comprises the following steps: step 1, remote wireless units of M antennas transmit 5GNR signals through a uniform linear array; 2, the remote wireless unit receives signals reflected back in the environment; step 3, clutter suppression is carried out on the received signal by using a multi-pulse cancellation algorithm; step 4, converting the signal obtained after clutter suppression into a composite model used for solving a sparse multi-measurement vector recovery problem; step 5, using a UAMP-SBL for MMV algorithm to solve the composite model, and identifying L rows of the matrix A corresponding to the L paths in the composite model; step 6, introducing all sensing parameters except time delay into the matrix A, and extracting non-zero lines in the matrix A to obtain a matrix A '; and step 7, performing sensing parameter estimation according to the matrix A '. The method not only minimizes clutter interference, but also reduces the number of required sensing parameters, and simplifies the subsequent sensing algorithm.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a method for clutter suppression and perception parameter estimation in a perception mobile network. Background Art

[0002] As mobile communication technology moves from the fifth generation (5G) to the sixth generation (6G), perceptual mobile networks (PMNs) have gradually become a research hotspot. Such networks not only have communication capabilities, but also can perceive the surrounding environment, realizing the integration of perception and communication (ISAC), and are widely used in fields such as autonomous driving, smart cities, and enhanced Internet of Things (IoT). However, the realization of such a complex network is accompanied by a series of challenges, especially in perception parameter estimation and clutter suppression. Clutter, that is, unwanted echoes from objects in the environment, can significantly degrade the performance of PMNs because it can mask the intended target and interfere with signal transmission. Clutter mainly refers to multipath signals from permanent or long-term static objects, which may significantly increase the number of perception parameters, so clutter suppression is necessary before parameter estimation.

[0003] There is little research work on clutter suppression techniques applied to mobile networks. An ISAC clutter estimation method based on maximum likelihood amplitude estimation uses a simple clutter model for sample average evaluation. However, for more complex clutter models, the residual of clutter cancellation may be higher, which in turn deteriorates the perceived performance. Taking advantage of the stability of the signal within the coherent time period, a recursive moving average (RMA) based method is proposed to suppress clutter by recursively averaging and smoothing the signal within a window and using a forgetting factor. Although this method effectively reduces clutter from static paths by filtering the signal within a fixed interval, its use of random and irregular demodulation reference signals (DMRS) in 5G-NR in the time domain makes it difficult to apply to PMNs.

[0004] In addition, there is a clutter estimation and suppression method based on Gaussian mixture model (GMM) and expectation maximization (EM), called GMM-EM-CE, which distinguishes static clutter from dynamic signals by statistically modeling the channel state using Gaussian distribution. Since the number of multipath components is unknown in the actual environment, it is difficult for this method to determine the number of Gaussian mixture components. In addition, this method requires matrix inversion in each cycle, resulting in high computational complexity. The above traditional methods indirectly require channel information for clutter suppression and then perform perceptual parameter estimation, which may introduce potential clutter and parameter estimation errors due to poor estimated channels. Therefore, it is necessary to design a clutter suppression method that does not rely on demodulation reference signals and has low complexity and direct estimation. Summary of the invention

[0005] In view of the deficiencies of the prior art, the present invention proposes a method for clutter suppression and perception parameter estimation in a perceptual mobile network, which not only minimizes clutter interference, but also reduces the number of required perception parameters and simplifies subsequent perception algorithms.

[0006] In order to solve the above technical problems, the technical solution of the present invention is:

[0007] A method for clutter suppression and perception parameter estimation in a perceptual mobile network comprises the following steps:

[0008] Step 1: A remote wireless unit with M antennas transmits a 5G NR signal through a uniform linear array;

[0009] Step 2: The remote wireless unit receives the signal reflected back from the environment;

[0010] Step 3: Use a multi-pulse cancellation algorithm to suppress clutter on the received signal;

[0011] Step 4: convert the signal obtained after clutter suppression into a composite model for solving the sparse multi-measurement vector recovery problem;

[0012] Step 5, applying the UAMP-SBL for MMV algorithm to solve the composite model, and identifying the L rows of the matrix A corresponding to the L paths in the composite model;

[0013] Step 6: Introduce all perceptual parameters except delay into matrix A, extract non-zero rows in matrix A to obtain matrix A′;

[0014] Step 7: Estimate the perception parameters based on the matrix A′.

[0015] The present invention adopts the downlink active perception scenario of the perception mobile network under the cloud radio access network (CRAN) configuration. Each remote radio unit (RRU) is configured with M antennas with a half-wavelength spacing. The channel model defines the delay, Doppler frequency and angle in multipath propagation, and these parameters remain unchanged within a channel static period. By converting the received signal into a frequency domain signal, a frequency domain channel matrix model is established.

[0016] Preferably, in step 2, the last three OFDM symbols of the SSB are selected for sensing, and the beamforming signal transmitted on the nth subcarrier of the kth OFDM symbol in the gth SSB is recorded as x n，t =w t s n，t , where g = 1, 2, ..., G, k = 1, 2, 3, w t is the beamforming vector of the tth OFDM symbol, where t = 4g + k, s n，t is a correlated phase modulated signal, and |sn，t | 2 =1, the signal received by the RRU on the nth subcarrier and the tth OFDM symbol is expressed as:

[0017]

[0018] where x n，t and z n，t They represent the transmitted signal vector and the noise vector respectively, θ is the departure angle or arrival angle, the departure angle and arrival angle of the target are the same, and the departure angle or arrival angle is recorded as θ l , l∈[1,L], where L is the number of paths, is the subcarrier spacing, N is the total number of subcarriers, B is the bandwidth, is the OFDM symbol period, T p represents the cyclic prefix period, a(M,θ l ) is the array response vector of RRU.

[0019] In the above technical solution, the synchronization signal fast SSB is used

[0020] Preferably, in step 3, the specific method of clutter suppression is:

[0021] The received OFDM symbols with the same index from P+1 consecutive SSB bursts are represented as Where N s is the number of OFDM symbols between two consecutive SSB burst sets. Directly processing the received signal according to the multi-pulse cancellation algorithm yields:

[0022]

[0023] (a) refers to the second equation in the formula, which is true because For SSB, the beamforming is the same and is the same. and is the frequency response of the clutter eliminator.

[0024] In the above technical solution, multipath clutter with zero Doppler shift is effectively eliminated through differential operation of the P frame SSB burst set.

[0025] Preferably, in step 3, when eliminating clutter, the parameter f D，l ≈0, and a larger P is chosen so that the clutter canceller has a deeper notch, thereby eliminating clutter close to zero Doppler frequency.

[0026] Preferably, the conversion method is:

[0027] Multiply the signal after clutter suppression by Get a signal model without data symbols:

[0028]

[0029] in Contains all perceptual parameters except delay, z′ n，t is the processed noise vector;

[0030] Then the signal models without data symbols of all N subcarriers are connected to obtain:

[0031] R t =[r 1，t , ..., r N,t ] T =CB t +Z′ t ,

[0032] where C = [c(τ1), ..., c(τ L )] contains the delay of all paths;

[0033] in and B t =[b 1，t , ..., b L，t ] T ;

[0034] Thus, the r of N subcarriers n，t Connect R t Re-expressed as a sparse model:

[0035] R t =C′IIB t +Z′ t ,

[0036] Where C′ is known and is quantized by the delay It is expressed as:

[0037] C′=[c(1 / (N d Δf)),...,c(N p / (N d Δf))],

[0038] Where N p Indicates the number of grids used, where L<<N p <N d ; Π is an N p ×L permutation matrix with L nonzero rows, each row containing a single nonzero element 1, t Mapped to the corresponding quantized delay,

[0039] The R corresponding to the three consecutive OFDM symbols of the g-th SSB t Connect and get:

[0040]

[0041] where t′=4g+1, and Represented by an N×3M observation matrix An N×N p The perception matrix C′ and an N p ×3M composed of sparse signal matrix;

[0042] All matrices Connect as:

[0043]

[0044] This results in a composite model:

[0045]

[0046] Preferably, in step 6, introducing And A′ is expressed as:

[0047]

[0048] Among them, each row of A′ corresponds to a path. If the g-th SSB beam is aligned with the l-th path, The power of the SSB will be higher than that of other SSBs.

[0049] Preferably, in step 7, when performing parameter estimation, it is necessary to determine the maximum Best g l ,Right now:

[0050]

[0051] Preferably, the perception parameter estimation method is:

[0052] pass and t′ l =4g l +1, you can get and

[0053]

[0054] Thus, the association {τ l , f D，l ,θ l , b lAll perceptual parameter estimates of}.

[0055] In the above technical solution, the sparse Bayesian learning (SBL) algorithm in the multiple measurement vector (MMV) problem is used to directly estimate the parameters of the received signal, avoiding the complex process of initial channel estimation. Through the sparse signal recovery problem model, the robustness and accuracy of the perception parameter estimation are improved, supporting high-precision and real-time processing requirements.

[0056] The present invention has the following characteristics and beneficial effects:

[0057] The present invention specifically addresses the unique needs of perceptual mobile networks (PMNs), solves the randomness and irregularity of demodulation reference signals (DMRS) in the time domain, and reduces the high complexity that is usually present in traditional clutter suppression methods. Through the innovative use of synchronization signal blocks (SSBs), the proposed method enhances clutter suppression capabilities.

[0058] In addition, a perception parameter algorithm is developed to complement the proposed clutter suppression method and further optimize the processing efficiency and effect in PMNs. This comprehensive approach not only minimizes clutter interference, but also reduces the number of required perception parameters and simplifies the subsequent perception algorithm.

[0059] The above technical solution can effectively suppress the clutter in the perception mobile network and improve the accuracy of perception parameter estimation; by directly processing the received signal, the calculation process is simplified and the calculation complexity is reduced; the sparse Bayesian learning algorithm is used to improve the robustness and real-time performance of perception parameter estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0061] Figure 1 The present invention provides a design process for the broadband hybrid array frequency domain compensation method.

[0062] Figure 2 It is a schematic diagram of distance and speed estimation of the present invention.

[0063] Figure 3 It is a schematic diagram of distance and arrival angle estimation of the present invention and the existing perception parameter estimation scheme.

[0064] Figure 4 It is the signal-to-noise ratio and the mean square error of the angle of arrival when the present invention is compared with the existing perception parameter estimation scheme. DETAILED DESCRIPTION

[0065] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0066] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings.

[0067] On the contrary, the present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention as defined by the claims. Further, in order to make the public have a better understanding of the present invention, some specific details are described in detail in the detailed description of the present invention below. Those skilled in the art can fully understand the present invention without the description of these details.

[0068] The present invention provides a method for clutter suppression and perception parameter estimation in a perceptual mobile network, such as Figure 1 As shown, the specific steps are as follows:

[0069] In this embodiment, a PMN based on a cloud radio access network (CRAN) is used, in which a remote radio unit (RRU) equipped with M antennas performs downlink active sensing through a uniform linear array (ULA) to estimate environmental parameters, that is, sensing using the echo of its transmitted signal. Define N as the total number of subcarriers, B as the bandwidth, and the subcarrier spacing is And the OFDM symbol period is Where T p represents the cyclic prefix period. Assuming plane wavefront signal propagation, the array response vector of RRU can be expressed as:

[0070] a(M,θ)=[1,e jπsin(θ) , ..., e jπ(M-1)sin(θ) ] H (1)

[0071] where θ is the angle of departure (AoD) or angle of arrival (AoA), and the antenna spacing is assumed to be half a wavelength.

[0072] Specifically, in this embodiment, a PMN with a single RRU equipped with four antennas is used, and its transmit power is set to 30 dBm. The carrier frequency and signal bandwidth are set to 2.35 GHz and 100 MHz respectively. The receiver thermal noise power is calculated by N0+10log(10 8 )=-94dBm, where the power spectrum density is N0=-174dBm / Hz. In the perception, an SSB signal with G=4 is used, which occupies 240 subcarriers and 12 symbols in the time-frequency domain. In the simulation, it is assumed that the radar cross-section of the target is 1 square meter.

[0073] It should be noted that for downlink active sensing, the target's AoD and AoA are the same. Therefore, AoDs(AoAs) are denoted as θ l , l∈[1,L], where L is the number of paths.

[0074] The M×M frequency domain channel matrix corresponding to the nth subcarrier of the tth OFDM symbol can be expressed as:

[0075]

[0076] where b l is the complex amplitude of the lth path, τ l is the propagation delay, f D，l is the corresponding Doppler frequency. In general, assuming τ l <τ l+1 The parameters that need to be estimated are {τ l , f D，l ,θ l , b l It is used to identify the position and velocity of the target. It is assumed that the wireless channel remains stationary for a few milliseconds, during which these perception parameters remain unchanged and the Doppler phase remains roughly constant at the sampling points in an OFDM symbol.

[0077] In order to achieve omnidirectional sensing, in this embodiment, the SSB signal of 5G NR is used for clutter suppression and sensing in PMN. In order to utilize as many subcarriers as possible, the last three OFDM symbols of SSB are selected for sensing. The beamforming signal transmitted on the nth subcarrier of the kth (k=1, 2, 3) OFDM symbol in the gth (g=1, 2, ..., G) SSB is x n，t =w t s n，t where w t is the beamforming vector of the tth OFDM symbol (t=4g+k), s n，t is a correlated phase modulated signal, and |s n，t | 2 =1, the signal received by the RRU on the nth subcarrier and the tth OFDM symbol can be expressed as:

[0078]

[0079] Among them, x n，t and z n，t represent the transmitted signal vector and the noise vector respectively.

[0080] In this embodiment, multipath signals are randomly generated in clusters to simulate reflected / scattered signals from objects. A large-scale path loss model is used with a path loss factor of 4. It is assumed that the number of multipaths, angular range, target distance, and Doppler frequency are uniformly distributed in the ranges of [15, 25], [0, 45] degrees, [0, 60] meters, and [0, 600] Hz, where the clutter signal includes a Doppler frequency close to zero. Additional offsets between the object and the RRU include an angular offset of [-75, 75] degrees, a distance offset of [60, 120] meters, and a moving speed offset of [-40, 40] meters / second. The delay is quantized to 10 nanoseconds, and the N d = 512 grids, which is equivalent to a distance quantization of 3 meters. Depending on the path loss factor and multipath propagation distance, the received signal-to-noise ratio (SNR) can be as low as 0 dB when estimating the sensing parameters.

[0081] A further arrangement of this embodiment, inspired by the moving target indicator (MTI) technology in radar, handles clutter by directly processing the received signal. This is in contrast to clutter suppression techniques in the literature, which require channel state information before clutter suppression. In particular, drawing on the multi-pulse canceller technology, the delay of the SSB signal repetition is used to eliminate the echo signal component from permanent or long-term stationary targets in PMNs. The received OFDM symbols with the same index from P+1 consecutive SSB bursts are represented as Where N s is the number of OFDM symbols between two consecutive SSB burst sets. According to the multi-pulse elimination principle, the received signal is directly processed and obtained:

[0082]

[0083] (a) holds because For SSB, the beamforming is the same and is the same. and

[0084]

[0085] Among them, a l It is the frequency response of the clutter eliminator. When P=1, it is a double pulse canceller, and when P=2, it is a triple pulse canceller, and so on. The brackets with big P and small p are the formulas of permutations and combinations.

[0086] It should be noted that the clutter eliminator is an algorithm for eliminating clutter from the received signal, so that subsequent perception can only perceive the moving object without being interfered by clutter. It is a conventional technical means and can be obtained directly, so it is not specifically described in this embodiment.

[0087] It can be understood that it can be easily verified from formula (2) that when f D,l = 0, a l = 0, so the clutter can be eliminated from equation (1). In order to deal with clutter with a Doppler frequency close to zero, that is, f D，l ≈0, a larger P needs to be chosen because the clutter canceller has a deeper notch. This will significantly reduce the number of parameters that need to be sensed and improve the sensing accuracy for dynamic multipaths or multipaths of interest with non-zero Doppler frequencies.

[0088] A further configuration of this embodiment is to convert the signal obtained after clutter suppression into a composite model for solving the problem of sparse multi-measurement vector recovery.

[0089] Specifically, multiply formula (1) by Get a signal model without data symbols:

[0090]

[0091] in Contains all perceptual parameters except delay, z′ n，t is the processed noise vector, It refers to the inverse of the transmitted signal. It can be understood that the transmitted signal is known, and the inverse of the transmitted signal and the transmitted signal cancel each other out.

[0092] Then r of all N subcarriers n，t Connecting them together, we get:

[0093] R t =[r 1，t , ..., r N，t ] T =CB t +Z′ t (4)

[0094] where C = [c(τ1), ..., c(τ L )] contains the delay of all paths, where and B t =[b 1，t , ..., b L，t ] T .

[0095] In order to facilitate delay estimation, the continuous delay terms Quantize to its nearest grid in Represents the quantized τ l Δf, grid size is N d >> L. It should be noted that broadband systems can usually provide sufficient delay resolution to meet the requirements of smaller delay quantization error.

[0096] In order to solve the problem of unknown matrix C, formula (4) is reformulated as a sparse model:

[0097] R t =C′∏B t +Z′ t (5)

[0098] Where C′ is known and is quantized by the delay It is expressed as:

[0099] C′=[c(1 / (N d Δf)),...,c(N p / (N d Δf))] (6)

[0100] Where N p (L<<N p <N d ) indicates the number of grids used.

[0101] Π is an N p ×L permutation matrix with L nonzero rows, each row containing a single nonzero element 1, t Mapped to the corresponding quantized delay. It should be noted that, in this embodiment, given R t and C′, we can solve the sparse signal recovery problem (5) to obtain the delay and IIB t Based on this, the remaining perceptual parameters can be effectively extracted.

[0102] It is understandable that traditional spectrum analysis and array signal processing techniques such as MUSIC and ESPRIT require continuous observation, which may not always be feasible in PMNs. Therefore, in this embodiment, compressed sensing (CS) is used to estimate perception parameters from complex and scattered signals, which has advantages in solving such problems.

[0103] First, sparse signal estimation is performed in formula (5) to obtain the delay estimation using the CS algorithm UAMP-SBL, thereby realizing the estimation and association of other perception parameters, where it is assumed that all paths have different delays.

[0104] The R corresponding to the three consecutive OFDM symbols of the g-th SSB t Connect as:

[0105]

[0106] where t′=4g+1, and

[0107] It can be seen that the problem in formula (7) is a sparse signal estimation problem with an N×3M observation matrix An N×N p The perception matrix C′ and an N p ×3M sparse signal matrix Only L rows are non-zero, that is, the matrix The sparse columns in share a common support set, which should be estimated Understandably, if The corresponding beam is not aimed at the target, and the power of its non-zero row may be small, so that the target may be missed. To solve this problem, in this embodiment, all matrices Connect as:

[0108]

[0109] This results in a composite model:

[0110]

[0111] It should be noted that since all column vectors in A share a common support set, it can be reformulated as a sparse MMV problem.

[0112] After applying the UAMP-SBL for MMV algorithm for sparse signal estimation, L rows corresponding to L paths can be identified in the matrix A. The non-zero rows in A are extracted to obtain A′, where l′ is the row index in the matrix A corresponding to the lth row in A′.

[0113] Specifically, in this embodiment, the sparse signal estimation method using the UAMP-SBL for MMV algorithm is as follows:

[0114]

[0115] In order to identify the beam aligned with the multipath, this embodiment introduces And A′ is expressed as:

[0116]

[0117] Each row of A′ corresponds to a path. If the g-th SSB beam is aligned with the l-th path, The power of the SSB will be higher than that of other SSBs.

[0118] In order to compare using any To improve the estimation accuracy, we need to determine the maximum Best g l ,Right now:

[0119]

[0120] in It refers to The second norm of .

[0121] pass and t′ l =4g l +1, get and

[0122]

[0123] as well as

[0124]

[0125] Thus, we estimate and relate {τ l , f D，l θ l , b l} all perceptual parameters.

[0126] In this embodiment, P=2 and recursive filtering are used to enhance the stability and accuracy of the perceived parameters over time. It iteratively optimizes the parameter estimates by combining new estimates, i.e. in and q(i) represent the filtered and new parameter estimation vectors of the ith SSB burst set, respectively, and α=0.9 is the forgetting factor. This method smoothes the noise and reduces the impact of abnormal measurements, which is particularly beneficial in dynamic environments.

[0127] Applying the above embodiment, Figure 2 The target range and velocity estimation performance using the proposed method is shown, highlighting that clutter with near-zero velocity can be effectively suppressed, while the target is preserved and the corresponding velocity is accurately estimated.

[0128] Figure 3 The proposed method, recursive moving average and Gaussian mixture model are compared in πsin(θ l ) in terms of AoA estimation. It can be seen from the results that the AoA estimation of the proposed method is more reliable and accurate than RMA+SBL and GMM-EM-CE+SBL regardless of the distance. This is attributed to the robustness and high performance of the UAMP algorithm. The algorithm avoids the divergence problem commonly seen in AMP by unit transformation of the measurement matrix.

[0129] In addition, the structured variational inference used in UAMP-SBL effectively exploits the sparse prior to recover the sparse signal. sThe probability characteristics of 10 iterations and 64 samples in a time interval of 10 are used to distinguish the target signal from the clutter. RMA takes advantage of the stability of the perception parameters in short time intervals and performs recursive averaging with a forgetting factor of 0.99 in 64 iterations.

[0130] Figure 4 The mean square error (MSE) of AoA estimation under different SNRs is compared. The results show that when the MSE is -20dB, the proposed method has 3dB and 5dB SNR gains over GMM-EM-CE+SBL and RMA+SBL, respectively. In addition, the performance of the proposed method is much better than that of UAMP-SBL without clutter suppression, which proves the necessity and effectiveness of the built-in clutter suppression in the proposed method.

[0131] The present invention studies the problem of perceptual parameter estimation in the presence of clutter in PMNs. The present invention proposes a novel low-complexity method by integrating a multi-pulse canceller-based clutter suppression mechanism into perceptual parameter estimation and using the UAMP algorithm for parameter association. In particular, the present invention exploits the beam scanning characteristics of 5G NR SSB signals. Simulation results show that the proposed method has advantages in perceptual performance and complexity compared with existing methods.

[0132] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions and variations of these embodiments including components are made without departing from the principles and spirit of the present invention, and still fall within the scope of protection of the present invention.

Claims

1. A method for clutter suppression and perception parameter estimation in a perceptual mobile network, characterized in that: The steps include: Step 1: A remote wireless unit with M antennas transmits a 5G NR signal through a uniform linear array; Step 2: The remote wireless unit receives the signal reflected back from the environment; Step 3: Use a multi-pulse cancellation algorithm to suppress clutter on the received signal; Step 4: convert the signal obtained after clutter suppression into a composite model for solving the sparse multi-measurement vector recovery problem; Step 5, using the UAMP-SBL for MMV algorithm to solve the composite model and identify the L rows of the matrix A corresponding to the L paths; Step 6: Introduce all perceptual parameters except delay into matrix A, extract non-zero rows in matrix A to obtain matrix A′; Step 7: Estimate the perception parameters based on the matrix A′.

2. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 1, characterized in that: In step 2, the last three OFDM symbols of the SSB are selected for sensing, and the beamforming signal transmitted on the nth subcarrier of the kth OFDM symbol in the gth SSB is recorded as x n，t =w t s n，t , where g = 1, 2, ..., G, k = 1, 2, 3, w t is the beamforming vector of the tth OFDM symbol, where t = 4g + k, s n，t is a correlated phase modulated signal, and |s n，t | 2 =1, the signal received by the RRU on the nth subcarrier and the tth OFDM symbol is expressed as: where x n，t and z n，t They represent the transmitted signal vector and the noise vector respectively, θ is the departure angle or arrival angle, the departure angle and arrival angle of the target are the same, and the departure angle or arrival angle is recorded as θ l , l∈[1,L], where L is the number of paths, is the subcarrier spacing, N is the total number of subcarriers, B is the bandwidth, is the OFDM symbol period, T p represents the cyclic prefix period, a(m,θ l ) is the array response vector of RRU.

3. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 2, characterized in that: In step 3, the specific method of clutter suppression is: Denote the received OFDM symbol with the same index from P+1 consecutive SSB bursts as y n，t ,..., Where N s is the number of OFDM symbols between two consecutive SSB burst sets. Directly processing the received signal according to the multi-pulse cancellation algorithm yields: (a) means that the equation holds because x n，t ,..., For SSB, the beamforming is the same and is the same. and is the frequency response of the clutter eliminator.

4. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 3, characterized in that: In step 3, when eliminating clutter, the parameter f D，l ≈0, and a larger P is chosen so that the clutter canceller has a deeper notch, thereby eliminating clutter close to zero Doppler frequency.

5. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 3, characterized in that: In step 4, the conversion method of the signal obtained after clutter suppression is: Multiply the signal after clutter suppression by Get a signal model without data symbols: in Contains all perceptual parameters except delay, z′ n，t is the processed noise vector, Refers to the inverse of the transmitted signal; Then the signal models without data symbols of all N subcarriers are connected to obtain: R t =[r 1，t ,...,r N，t ] T =CB t +Z′ t , where C = [c(τ1), ..., c(τ L )] contains the delay of all paths; in and B t =[b 1，t , ..., b L，t ] T ; Thus, the r of N subcarriers n，t Connect R t Re-expressed as a sparse model: R t =C′IIB t +Z′ t , Where C′ is known and is quantized by the delay It is expressed as: C′=[c(1 / (N d Δf)),...,c(N p / (N d Δf))], Where N p Indicates the number of grids used, where L<<N p <<N d ; Π is an N p ×L permutation matrix with L nonzero rows, each row containing a single nonzero element 1, t Mapped to the corresponding quantized delay, The R corresponding to the three consecutive OFDM symbols of the g-th SSB t Connect and get: where t′=4g+1, and Represented by an N×3M observation matrix An N×N p The perception matrix C′ and an N p ×3M composed of sparse signal matrix; All matrices Connect as: This results in a composite model:

6. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 5, characterized in that: In step 6, introduce And A′ is expressed as: Among them, each row of A′ corresponds to a path. If the g-th SSB beam is aligned with the l-th path, The power of the SSB will be higher than that of other SSBs.

7. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 6, characterized in that: In step 7, when performing parameter estimation, it is necessary to determine the maximum Best g l ,Right now:

8. The method for clutter suppression and perception parameter estimation in a perceptual mobile network according to claim 7, characterized in that: The estimation method of the perception parameters is: pass and t′ l =4g l +1, get and Thus, the association {τ l , f D，l ,θ l , b l All perceptual parameter estimates of}.

9. A method for clutter suppression and perception parameter estimation in a perceptual mobile network according to any one of claims 1 to 8, characterized in that: The antenna spacing in the remote wireless unit is half a wavelength.

Images