MIMO-OTFS system channel estimation method based on radar perception assistance and atom threshold

Abstract

The invention discloses an MIMO-OTFS system channel estimation method based on radar perception assistance and an atom selection threshold, and belongs to the technical field of wireless communication. According to the method, a channel is modeled based on a scene of a downlink of an MIMO-OTFS system, and a channel estimation problem is converted into a sparse channel matrix recovery problem in combination with a compressed sensing theory. A radar signal receiving and transmitting system is added to the base station and used for obtaining sensing information of the UE and the surrounding environment; and initializing the index of the angle domain OTFS channel according to the UE propagation path information. Sparse channel estimation of the MIMO-OTFS system is improved by using perception information extracted by a radar, and an MIMO-OTFS channel estimation algorithm based on perception assistance and an atom selection threshold is constructed. According to the channel estimation algorithm provided by the invention, the estimation precision of the wireless channel can be improved, the pilot frequency overhead of channel estimation is reduced, and the method has a certain application value in a radar perception assisted MIMO-OTFS wireless communication system.

Description

Technical Field

[0001] The invention relates to the technical field of wireless communication, and in particular to a MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold. Background Art

[0002] With the development of 5G mobile communication systems, academia and industry have launched research on 6G. 6G will have more detailed scenarios and need to meet multi-dimensional extreme performance requirements. To this end, it is necessary to deeply integrate traditional wireless perception functions such as positioning, detection, and imaging with wireless transmission functions, and use widely distributed computing power for auxiliary calculations to achieve cross-integration of perception, communication, and calculations. The reliability of communication in high-mobility scenarios is closely related to the performance of channel estimation. Robust and accurate channel estimation algorithms can improve the reliability of communication in this scenario. In order to achieve high-mobility communication, higher reliability and higher data rates are required. In communication systems based on orthogonal frequency-division multiplexing (OFDM) technology, the fast time-varying characteristics of high-mobility scenario channels and intercarrier interference (ICI) caused by Doppler spread pose challenges to the reliability of communication. To address this problem, the orthogonal time frequency space (OTFS) modulation scheme was proposed. The main innovation of OTFS is to implement information symbol multiplexing on a two-dimensional orthogonal basis function in the delay-Doppler (DD) domain. This orthogonal basis function can be used to combat the dynamic characteristics of time-varying multipath channels. Therefore, OTFS converts fading, time-varying multipath channels into sparse and slowly time-varying channels.

[0003] Based on the sparse characteristics of the DD domain channel, the sparse signal recovery method can be used for channel estimation. For the multiple-input multiple-output OTFS (MIMO-OTFS) system, existing research uses the three-dimensional sparsity of the channel in the delay-Doppler angle domain (DDA) to transform the downlink channel estimation problem into a sparse signal recovery problem, and solves the problem by iteratively estimating the three-dimensional support set of each propagation path, which can obtain accurate channel state information with low pilot overhead; using the sparsity of the channel in DDA, through tensor decomposition and parallel support set detection, a clipping operation is performed in the OMP algorithm to remove the support set that contributes less to the residual during the iteration process, reducing the computational complexity of channel estimation.

[0004] While the above methods use the sparsity of DD domain channels to improve channel estimation accuracy, they are also accompanied by high algorithm complexity and computational overhead due to the large number of iterations. In high mobility scenarios, too many iterations of the algorithm will cause delays in updating the channel state information, which cannot truly reflect the actual situation of the current channel and affect the performance of the system channel estimation. On the other hand, in the MIMO-OTFS system, the increase in the number of antennas increases the computational overhead of the estimation algorithm, and the maximum delay and Doppler spread reduce the estimation accuracy. Summary of the invention

[0005] In view of the above problems, the present invention proposes a MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold to improve the channel estimation performance.

[0006] The technical solution of the present invention is: a MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold, the specific steps are:

[0007] Step 1: Based on the radar perception-assisted MIMO-OTFS system downlink scenario, a channel model is established to derive the relationship between the received signal at the UE, the base station transmitted signal and the channel.

[0008] Step 1.1: Consider a radar-assisted downlink MIMO-OTFS system. The base station is equipped with a linear frequency modulated wave (FMCW) radar and uses N T The OTFS transmitter with a uniform linear array (ULA) of 10 antennas is transmitted at a carrier frequency f c Communicate with high-speed mobile user equipment (UE) configured with a single antenna.

[0009] Step 1.2: Consider a time-varying channel model with L propagation paths and calculate the delay tap index m of the DD domain signal of the i-th path i and Doppler tap index n i :

[0010] m i =round(MΔfτ i )

[0011] n i =round(NTν i )

[0012] In the formula, τ i , ν i They represent the delay and angle of departure (AoD) of the i-th (i∈[1,…,L]) path, M represents the delay dimension of the DD domain signal, N represents the Doppler dimension of the DD domain signal, MΔf represents the bandwidth of the OTFS system, and NT represents the duration of an OTFS frame.

[0013] Step 1.3: The channel from the ath transmitting antenna to the UE can be expressed as:

[0014]

[0015] Where: α i , i They represent the complex gain and departure angle of the i-th path respectively; m is the index of the delay tap; n is the index of the Doppler tap; d and λ are the antenna spacing and wavelength of the OTFS transmitter respectively.

[0016] Step 1.4: Received signal at UE:

[0017]

[0018] Where: M g is the guard interval of the transmitted DD domain signal delay dimension, z is the phase compensation, is the delayed Doppler signal X sent by the ath antenna DD The element in the m′th row and n′th column. is the noise of the DD domain signal.

[0019] Step 2: Combine the compressed sensing theory to perform secondary modeling on the channel model. Convert the channel estimation problem into a sparse channel matrix recovery problem.

[0020] Step 2.1: Write the received signal at the UE in matrix-vector form:

[0021] y=(Z⊙P)h+v

[0022] Where: and are the received signal and the channel from the ath transmitting antenna to the UE respectively The vector form of and are the phase compensation matrix and the pilot signal matrix respectively, M pl is the pilot symbol length of the transmitted DD domain signal delay dimension.

[0023] Step 2.2: Write the received signal as the product of the sensing matrix and the angle domain sparse channel to solve the channel estimation problem using the compressed sensing algorithm:

[0024]

[0025] Where: I is the identity matrix and F is the Fourier transform matrix.

[0026] Step 3: Establish a radar signal transceiver system at the base station to obtain sensing information of the UE and the surrounding environment.

[0027] Step 3.1: The radar emits N loop The same linear frequency modulation signals are used to form radar frames. The transmission signal in a radar frame can be expressed as:

[0028]

[0029] Where: s chirp (t) is the linear frequency modulation signal; T p is the repetition period of the linear frequency modulation pulse; T f is the radar frame duration.

[0030] Step 3.2: Use a quadrature mixer to mix the received signal at each antenna with the transmitted signal to obtain in-phase and quadrature components. The in-phase and quadrature components are passed through a low-pass filter to obtain an intermediate frequency (IF) signal:

[0031]

[0032] Where: β is the complex gain of the radar propagation path; b is the number of radar transmitting antennas; f 0 The starting frequency of the linear FM signal, is the round-trip propagation delay, D w is the propagation distance between the radar and the wth ideal reflector, c is the speed of light; d r and λ r is the wavelength and antenna spacing of the radar.

[0033] Step 3.3: Sample the intermediate frequency signal to form a radar data frame X, and use the mean cancellation algorithm to remove clutter from the radar data frame X to filter the reflected signal of the static object:

[0034]

[0035] Where: X r is the radar data frame after clutter removal, k is the sampling point of the delay dimension of the radar data frame, and l is the sampling point of the Doppler dimension of the radar data frame;

[0036] Step 3.4: In X r 3D-DFT is applied to extract the distance, Doppler and angle information corresponding to the moving target in the radar data frame to form the radar 3D thermal spectrum X 3D :

[0037] X 3D =|DFT 3D(X r )|

[0038] The constant false alarm rate (CFAR) detection algorithm is used to detect X 3D After peak detection, the propagation delay, Doppler velocity and angle of arrival (AoA) of each peak are extracted to represent the UE propagation path information:

[0039]

[0040] Where J is the number of peaks detected from radar peak detection, represents the propagation delay, Doppler velocity and AoA corresponding to the j{j=[1,…,J]}th peak.

[0041] Step 4: According to the propagation delay, Doppler speed and AoA of each peak in the UE propagation path information, Convert to angle domain OTFS channel The index of .

[0042] Step 4.1: Normalize the propagation delay of the propagation path detected by the radar:

[0043]

[0044] Step 4.2: Multiply the Doppler velocity by the carrier frequency of the communication system to convert each detected peak Doppler velocity into a Doppler frequency:

[0045]

[0046] Step 4.3: Normalize the propagation delay and Doppler frequency Converted to the delay tap and Doppler tap index of the DD domain channel respectively and

[0047] Step 4.4: The AoA peak value detected by the radar Convert to The row index f j :

[0048]

[0049] Step 4.5: Convert UE propagation path information into perception support set S r ={t 1 ,…,t j}, where t j for Corresponding to The index of the jth detected peak in tj It can be expressed as

[0050]

[0051] Step 5: Sent to UE, where the radar is used to extract the sensing support set S r To improve MIMO-OTFS sparse channel estimation, we build a channel estimation algorithm:

[0052] Step 5.1: Set the algorithm input: receive pilot signal y, perception matrix Perception Information Path threshold ρ, the number of atoms P selected in each iteration;

[0053] Step 5.2: Initialize variables and constants: channel estimation Residual Iteration number t = 1, support set Path threshold ρ = 2;

[0054] Step 5.3: When

[0055] Step 5.4: Select the inner product The largest P value e P and the corresponding index {φ(i)} i＝1,2,…,P ;

[0056] Step 5.5: Calculate the secondary screening threshold Where: u t1 and u t2 are the first two largest values ​​of the inner product e of the perception matrix and the residual; μ is the compromise factor.

[0057] Step 5.6: Select e P Zhonge P ≥χ (t) P χ Values ​​and corresponding indices

[0058] Step 5.7: Update support set:

[0059] Step 5.8: Perform least squares solution:

[0060] Step 5.9: If choose The largest The index corresponding to the value, update and

[0061] Step 5.10: Otherwise, update the residual:

[0062] Step 5.11: When Return to Step 5.3;

[0063] Step5.12: Otherwise output:

[0064] The beneficial effects of the present invention are:

[0065] Aiming at the channel estimation problem of MIMO-OTFS system, the present invention combines the compressed sensing theory to transform the channel estimation problem into the sparse channel matrix recovery problem. Based on the generalized orthogonal matching pursuit (GOMP) algorithm, the GOMP algorithm is improved by introducing radar perception system, atom selection threshold and iterative stop condition based on the number of peak detections, and a GOMP channel estimation algorithm based on perception assistance and atomic threshold is proposed, named SA-TGOMP.

[0066] The SA-TGOMP proposed in the present invention improves GOMP from multiple angles. First, the present invention proposes to obtain the sensing information of UE and the surrounding environment by introducing a radar sensing system, and initialize the index of the downlink DDA domain channel according to the UE propagation path information, so as to improve the sparse channel estimation accuracy of the MIMO-OTFS system by using the downlink sensing information extracted by the radar. Secondly, the present invention proposes an atom selection threshold, and uses the atom selection threshold to perform secondary screening on the atoms selected by the GOMP algorithm, and screens the atoms with the highest similarity to the current residual to update the support set, which improves the reliability of the algorithm to update the support set, thereby enhancing the estimation performance. Finally, the iteration stop condition based on the number of peak detections sets the number of algorithm iterations adaptively according to the path information sensed by the radar, reduces the search space of the algorithm, and improves the overall performance of the algorithm. The experimental results show that the proposed algorithm can effectively identify the MIMO-OTFS system channel, and has obvious advantages over the comparative algorithm in terms of estimation accuracy and pilot overhead. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a model diagram of a radar perception-assisted MIMO-OTFS communication system according to an embodiment of the present invention;

[0068] Figure 2 A model diagram of a radar signal processing system according to an embodiment of the present invention;

[0069] Figure 3 This is a comparison chart of the NMSE performance of the present invention and the existing solution under different SNRs of the present invention;

[0070] Figure 4 The figure is a comparison chart of NMSE performance between the present invention and the existing solution under different pilot overheads of the present invention. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0072] A MIMO-OTFS system channel estimation method based on radar perception assistance and atomic thresholding specifically comprises the following steps:

[0073] Step 1: Based on the radar perception-assisted MIMO-OTFS system downlink scenario, a channel model is established to derive the relationship between the received signal at the UE, the base station transmitted signal and the channel.

[0074] Step 1.1: Figure 1 As shown in FIG. 1 , a system model diagram of the present invention is shown. Consider a radar-assisted downlink MIMO-OTFS system. The base station is equipped with a linear frequency modulated wave (FMCW) radar and uses N T The OTFS transmitter with a uniform linear array (ULA) of 10 antennas is transmitted at a carrier frequency f c Communicate with high-speed mobile user equipment (UE) configured with a single antenna.

[0075] Step 1.2: Consider a time-varying channel model with L propagation paths and calculate the delay tap index m of the DD domain signal of the i-th path i and Doppler tap index n i :

[0076] m i =round(MΔfτ i )

[0077] n i =round(NTν i )

[0078] In the formula, τ i , ν i They represent the delay and angle of departure (AoD) of the i-th (i∈[1,…,L]) path, M represents the delay dimension of the DD domain signal, N represents the Doppler dimension of the DD domain signal, MΔf represents the bandwidth of the OTFS system, and NT represents the duration of an OTFS frame.

[0079] Step 1.3: The channel from the ath transmitting antenna to the UE can be expressed as:

[0080]

[0081] Where: α i , i They represent the complex gain and departure angle of the i-th path respectively; m is the index of the delay tap; n is the index of the Doppler tap; d and λ are the antenna spacing and wavelength of the OTFS transmitter respectively.

[0082] Step 1.4: Received signal at UE:

[0083]

[0084] Where: M g is the guard interval of the transmitted DD domain signal delay dimension, z is the phase compensation, is the delayed Doppler signal X sent by the ath antenna DD The element in the m′th row and n′th column. is the noise of the DD domain signal.

[0085] Step 2: Combine the compressed sensing theory to perform secondary modeling on the channel model. Convert the channel estimation problem into a sparse channel matrix recovery problem.

[0086] Step 2.1: Write the received signal at the UE in matrix-vector form:

[0087] y=(Z⊙P)h+v

[0088] Where: and are the received signal and the channel from the ath transmitting antenna to the UE respectively The vector form of and are the phase compensation matrix and the pilot signal matrix respectively, M pl is the pilot symbol length of the transmitted DD domain signal delay dimension.

[0089] Step 2.2: Write the received signal as the product of the sensing matrix and the angle domain sparse channel to solve the channel estimation problem using the compressed sensing algorithm:

[0090]

[0091] Where: I is the identity matrix and F is the Fourier transform matrix.

[0092] Step 3: Figure 2 As shown, it is a system model diagram of radar signal processing of the present invention. A radar signal transceiver system is established at the base station to obtain sensing information of the UE and the surrounding environment.

[0093] Step 3.1: The radar emits Nloop The same linear frequency modulation signals are used to form radar frames. The transmission signal in a radar frame can be expressed as:

[0094]

[0095] Where: s chirp (t) is the linear frequency modulation signal; T p is the repetition period of the linear frequency modulation pulse; T f is the radar frame duration.

[0096] Step 3.2: Use a quadrature mixer to mix the received signal at each antenna with the transmitted signal to obtain in-phase and quadrature components. The in-phase and quadrature components are passed through a low-pass filter to obtain an intermediate frequency (IF) signal:

[0097]

[0098] Where: β is the complex gain of the radar propagation path; b is the number of radar transmitting antennas; f 0 The starting frequency of the linear FM signal, is the round-trip propagation delay, D w is the propagation distance between the radar and the wth ideal reflector, c is the speed of light; d r and λ r is the wavelength and antenna spacing of the radar.

[0099] Step 3.3: Sample the intermediate frequency signal to form a radar data frame X, and use the mean cancellation algorithm to remove clutter from the radar data frame X to filter the reflected signal of the static object:

[0100]

[0101] Where: X r is the radar data frame after clutter removal, k is the sampling point of the delay dimension of the radar data frame, and l is the sampling point of the Doppler dimension of the radar data frame;

[0102] Step 3.4: In X r 3D-DFT is applied to extract the distance, Doppler and angle information corresponding to the moving target in the radar data frame to form the radar 3D thermal spectrum X 3D :

[0103] X 3D =|DFT 3D (X r )|

[0104] The constant false alarm rate (CFAR) detection algorithm is used to detect X 3DAfter peak detection, the propagation delay, Doppler velocity and angle of arrival (AoA) of each peak are extracted to represent the UE propagation path information:

[0105]

[0106] Where J is the number of peaks detected from radar peak detection, represents the propagation delay, Doppler velocity and AoA corresponding to the j{j=[1,…,J]}th peak.

[0107] Step 4: According to the propagation delay, Doppler speed and AoA of each peak in the UE propagation path information, Convert to angle domain OTFS channel The index of .

[0108] Step 4.1: Normalize the propagation delay of the propagation path detected by the radar:

[0109]

[0110] Step 4.2: Multiply the Doppler velocity by the carrier frequency of the communication system to convert each detected peak Doppler velocity into a Doppler frequency:

[0111]

[0112] Step 4.3: Normalize the propagation delay and Doppler frequency Converted to the delay tap and Doppler tap index of the DD domain channel respectively and

[0113] Step 4.4: The AoA peak value detected by the radar Convert to The row index f j :

[0114]

[0115] Step 4.5: Convert UE propagation path information into perception support set S r ={t 1 ,…,t j}, where t j for Corresponding to The index of the jth detected peak in t j It can be expressed as

[0116]

[0117] Step 5: Sent to UE, where the radar is used to extract the sensing support set S r To improve MIMO-OTFS sparse channel estimation, we build a channel estimation algorithm:

[0118] Step 5.1: Set the algorithm input: receive pilot signal y, perception matrix Perception Information Path threshold ρ, the number of atoms P selected in each iteration;

[0119] Step 5.2: Initialize variables and constants: channel estimation Residual Iteration number t = 1, support set Path threshold ρ = 2;

[0120] Step 5.3: When

[0121] Step 5.4: Select the inner product The largest P value e P and the corresponding index {φ(i)} i＝1,2,…,P ;

[0122] Step 5.5: Calculate the secondary screening threshold Where: u t1 and u t2 are the first two largest values ​​of the inner product e of the perception matrix and the residual; μ is the compromise factor.

[0123] Step 5.6: Select e P Zhonge P ≥χ (t) P χ Values ​​and corresponding indices

[0124] Step 5.7: Update support set:

[0125] Step 5.8: Perform least squares solution:

[0126] Step 5.9: If choose The largest The index corresponding to the value, update and

[0127] Step 5.10: Otherwise, update the residual:

[0128] Step 5.11: When Return to Step 5.3;

[0129] Step5.12: Otherwise output:

[0130] In order to verify the performance of the method of the present invention, software simulation was performed using Matlab. A channel model based on 3D ray tracing was used, and UEs were randomly distributed in a section of a city street, and their random speeds were uniformly sampled within [50,90] m / s. 1000 scenarios with different user positions and speeds were sampled. The complex gain, propagation delay, Doppler frequency, AoA and AoD parameters of each propagation path were simulated for each scenario, and the communication channel was obtained based on these channel parameters. and radar data frame X. In the OTFS transmitter parameters: the number of transmitting antennas is 32, the number of receiving antennas is 1, the carrier frequency is 6 GHz, the number of subcarriers is 256, the number of symbols is 14, and the subcarrier spacing is 15 kHz. In the radar parameters: the starting frequency is 28 GHz, the number of linear frequency modulation signals is 256, the number of radar transmitting antennas is 1, and the number of radar receiving antennas is 16. The results of the above specific embodiments are as follows Figure 3 As shown, it is a comparison chart of the NMSE performance of the present invention and the existing scheme under different SNRs. The SA-TGOMP estimation algorithm of the present invention has the best NMSE performance. Compared with other compressed sensing algorithms that directly use the sparsity of the DD domain channel, the algorithm of the present invention uses radar sensing information to initialize the estimated support set, and the introduction of the atom selection threshold improves the atomic correlation of the support set. Therefore, SA-TGOMP greatly improves the estimation performance. Compared with the SA-OMP algorithm, the NMSE of the algorithm in this paper is reduced by about 2dB.

[0131] Figure 4 The NMSE performance of the proposed method and the existing scheme under different pilot overheads is evaluated. The pilot overhead η∝C log(N T M g N), C is the channel sparsity level, N T M g N is the length of the sparse channel h in the DD domain. When the number of antennas increases, the pilot demand for channel estimation increases and shows a convergence trend. From the experimental results, it can be observed that when more pilot signals are used for channel estimation, the NMSE performance improves, and the performance increase decreases with each increase in pilot overhead. When a lower pilot overhead η is used for estimation, the proposed SA-TGOMP algorithm has a significant improvement in NMSE performance compared to the SA-OMP algorithm; when a higher pilot overhead ratio is used for estimation, the proposed SA-TGOMP algorithm has a smaller improvement in NMSE performance than SA-OMP. This shows that the SA-TGOMP algorithm can achieve similar or even stronger estimation performance than other methods without increasing pilot signals.

[0132] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold, characterized in that: The following steps are involved: Step 1: Based on the radar perception-assisted MIMO-OTFS system downlink scenario, a channel model is established to derive the relationship between the received signal at the UE, the base station transmitted signal and the channel; Step 2: Combined with the compressed sensing theory, the channel model is remodeled to transform the channel estimation problem into a sparse channel matrix recovery problem; Step 3: Establish a radar signal transceiver system at the base station to obtain sensing information of the UE and the surrounding environment; Step 4: Convert the propagation delay, Doppler velocity and AoA of each peak in the UE propagation path information into the index of the OTFS channel in the angle domain; Step 5: Send the sensing information obtained by radar processing to the UE, where the perception support set extracted by the radar is used to improve the MIMO-OTFS sparse channel estimation and construct a channel estimation method.

2. The MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold according to claim 1, characterized in that: The Step 1 is specifically as follows: Step 1.1: Consider a radar-assisted downlink MIMO-OTFS system. The base station is equipped with a linear frequency modulation radar and uses N T The OTFS transmitter with a uniform linear array of antennas is c Communicate with high-speed mobile UEs equipped with a single antenna; Step 1.2: Consider a time-varying channel model with L propagation paths and calculate the delay tap index m of the DD domain signal of the i-th path i and Doppler tap index n i : m i =round(MΔfτ i ) n i =round(NTν i ) In the formula, τ i , ν i They represent the delay and departure angle of the i-th path, respectively, M represents the delay dimension of the DD domain signal, N represents the Doppler dimension of the DD domain signal, MΔf represents the bandwidth of the OTFS system, and NT represents the duration of an OTFS frame; Step 1.3: The channel from the ath transmitting antenna to the UE can be expressed as: Where: α i , i They represent the complex gain and departure angle of the ith path, respectively; m is the index of the delay tap; n is the index of the Doppler tap; d and λ are the antenna spacing and wavelength of the OTFS transmitter, respectively; Step 1.4: Received signal at UE: Where: M g is the guard interval of the transmitted DD domain signal delay dimension, z is the phase compensation, is the delayed Doppler signal X sent by the ath antenna DD The element in the m′th row and n′th column, is the noise of the DD domain signal.

3. The MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold according to claim 1 is characterized in that: The Step 2 is specifically as follows: Step 2.1: Write the received signal at the UE in matrix-vector form: y=(Z⊙P)h+v Where: and are the received signal and the channel from the ath transmitting antenna to the UE respectively The vector form of and are the phase compensation matrix and the pilot signal matrix respectively, M pl is the pilot symbol length of the transmitted DD domain signal delay dimension; Step 2.2: Write the received signal as the product of the sensing matrix and the angle domain sparse channel to solve the channel estimation problem using the compressed sensing algorithm: Where: I is the identity matrix and F is the Fourier transform matrix.

4. The MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold according to claim 1, characterized in that: The Step 3 is specifically as follows: Step 3.1: The radar emits N loop The same linear frequency modulation signals are used to form radar frames. The transmission signal in a radar frame can be expressed as: Where: s chirp (t) is the linear frequency modulation signal; T p is the repetition period of the linear frequency modulation pulse; T f is the radar frame duration; Step 3.2: Use an orthogonal mixer to mix the received signal at each antenna with the transmitted signal to obtain in-phase and quadrature components. The in-phase and quadrature components are passed through a low-pass filter to obtain an intermediate frequency signal: Where: β is the complex gain of the radar propagation path; b is the number of radar transmitting antennas; f0 is the starting frequency of the linear frequency modulation signal, is the round-trip propagation delay, D w is the propagation distance between the radar and the wth ideal reflector, c is the speed of light; d r and λ r is the wavelength and antenna spacing of the radar; Step 3.3: Sample the intermediate frequency signal to form a radar data frame X, and use the mean cancellation algorithm to remove clutter from the radar data frame X to filter the reflection signal of the static object: Where: X r is the radar data frame after clutter removal, k is the sampling point of the delay dimension of the radar data frame, and l is the sampling point of the Doppler dimension of the radar data frame; Step 3.4: In X r 3D-DFT is applied to extract the distance, Doppler and angle information corresponding to the moving target in the radar data frame to form the radar 3D thermal spectrum X 3D : X 3D =|DFT 3D (X r )| Step 3.5: Use CFAR detection algorithm to detect X 3D After peak detection, the propagation delay, Doppler speed and AoA of each peak are extracted to represent the UE propagation path information: Where J is the number of peaks detected from radar peak detection, represents the propagation delay, Doppler velocity and AoA corresponding to the j{j=[1,…,J]}th peak.

5. The MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold according to claim 1, characterized in that: The Step 4 is specifically as follows: Step 4.1: Normalize the propagation delay of the propagation path detected by the radar: Step 4.2: Multiply the Doppler velocity by the carrier frequency of the communication system to convert each detected peak Doppler velocity into a Doppler frequency: Step 4.3: Normalize the propagation delay and Doppler frequency Converted to the delay tap and Doppler tap index of the DD domain channel respectively and Step 4.4: The AoA peak value detected by the radar Convert to The row index f j : Step 4.5: Convert UE propagation path information into perception support set S r ={t1,…,t j }, where t j for Corresponding to The index of the jth detected peak in t j It can be expressed as:

6. The MIMO-OTFS system channel estimation method based on radar perception assistance and atomic threshold according to claim 1, characterized in that: The Step 5 is specifically as follows: Step 5.1: Set the algorithm input: receive pilot signal y, perception matrix Perception Information Path threshold ρ, the number of atoms P selected in each iteration; Step 5.2: Initialize variables and constants: channel estimation Residual Iteration number t = 1, support set Path threshold ρ = 2; Step 5.3: When Step 5.4: Select the inner product The largest P value e P and the corresponding index {φ(i)} i＝1,2,…,P ; Step 5.5: Calculate the secondary screening threshold Where: u t1 and u t2 are the first two largest values ​​of the inner product e of the perception matrix and the residual, and μ is the compromise factor; Step 5.6: Select e P Zhonge P ≥χ (t) P χ Values ​​and corresponding indices Step 5.7: Update support set: Step 5.8: Perform least squares solution: Step 5.9: If choose The largest The index corresponding to the value, update and Step 5.10: Otherwise, update the residual: Step 5.11: When Return to Step 5.3; Step5.12: Otherwise output:

Images