Method for realizing perception based on SSB signal in 5G NR

Abstract

The invention discloses a method for realizing perception based on an SSB signal in 5G NR, and the method comprises the following steps: an RRU periodically transmits a scanning signal of the SSB, and meanwhile, a user and a target in an environment reflect the transmitted signal; the RRU receives the reflected signals, data in the reflected signals are stripped and written into a compressed sensing form, the data are converted into data delayed in a grid model, unitary transformation is carried out, and then estimation is carried out through a UAMP-SBL algorithm; writing an obtained multipath estimation result into a compressed sensing form according to the condition that multiple paths have the same delay, and separating and estimating the multiple paths at different angles by using an OGSBI algorithm; and judging which RRU each group of parameters belongs to according to the estimated angle and the emission angles of different RRUs during emission. According to the method, the sensing precision can be remarkably improved, the calculation complexity is reduced, and higher robustness is achieved.

Description

Technical Field

[0001] The present invention relates to the field of parameter perception technology, and specifically to a method for realizing perception based on an SSB signal in 5G NR. Background Art

[0002] With the large-scale commercial deployment of fifth-generation (5G) mobile networks, the global telecommunications industry has quickly begun research on the sixth-generation (6G) mobile technology. This forward momentum aims to usher in a new era characterized by ubiquitous intelligent connectivity. To redefine connectivity, future wireless networks are expected to support many advanced applications that require high-performance communication and perception capabilities. These applications include network-enabled robotics, immersive virtual reality, and vehicle-to-vehicle communication systems, which have driven further research and development on integrated sensing and communication (ISAC). The integration of ISAC enables communication and perception to utilize common signal transmission and largely share hardware and signal processing components, allowing vehicle networks to perform functions such as environmental perception, object detection, and collision avoidance.

[0003] ISAC applications in mobile or cellular networks have become an important research area, which uses communication waveforms to extract perception information through target echoes. Due to the wide broadband coverage and powerful infrastructure of mobile networks, mobile networks with perception capabilities have the potential to transform into ubiquitous wireless sensors while continuing to provide communication services. In view of the various advantages of ISAC in mobile networks, related studies have proposed a framework for perceptual mobile networks (PMNs), which use communication-centric waveforms for perception and can perceive and understand the mobile environment in real time, thereby supporting various intelligent applications and services.

[0004] In PMN, there are two typical sensing methods: uplink sensing and downlink sensing. For uplink sensing, the uplink communication signal of the user equipment (UE) is used in the remote radio unit (RRU) to sense the environment between the UE and the RRU. The challenge of uplink sensing is that the clocks between the UE transmitter and the RRU receiver are not synchronized, which leads to parameter estimation ambiguity. To solve this problem, a related study proposed a sensing scheme based on cross-antenna cross correlation (CACC), which eliminates the ambiguity caused by time and carrier frequency offset and can accurately estimate multiple targets.

[0005] Downlink perception involves estimating environmental parameters using downlink communication signals transmitted from one or more RRUs. Related studies have developed an indirect perception parameter estimation scheme through signal stripping technology, which uses traditional channel estimation methods to extract multi-user signals from received signals by removing known data symbols. Although these indirect methods can obtain perception parameters and provide flexibility for perception, they have high computational complexity due to the matrix inversion operations involved. In addition, significant noise enhancement may lead to serious estimation errors. There are also studies that use reflected signals for perception estimation through direct estimation methods. However, this estimation method introduces very high complexity and is greatly limited in practical applications.

[0006] Despite these advances, there are still challenges in integrating these sensing methods with the existing 5G New Radio (NR) protocol. One of the main issues is that the communication signals of the existing 5G NR protocol cannot be directly used to implement sensing functions. This limitation highlights the need for specialized adjustments and poses challenges to the seamless integration of PMNs in the current 5G infrastructure. In addition, the additional time and frequency resources required for environmental sensing reduce the communication efficiency of PMNs, and a balance needs to be found between communication and sensing functions. To balance the needs of communication and sensing, the number and position of DMRS OFDM symbols need to be adjusted in the resource grid, which further poses challenges to resource optimization. These all require innovative solutions that can dynamically adapt to changing conditions without affecting network performance or service quality. Summary of the invention

[0007] The purpose of the present invention is to overcome the shortcomings of the prior art and propose a method for realizing perception based on the SSB signal in 5G NR. The method perceives the environment around multiple remote radio units (RRUs) based on the existing synchronization signal blocks (SSBs) in 5G NR. The method combines the two modes of downlink active perception and passive perception, which can effectively realize environmental perception without affecting communication, and provide support for various intelligent applications and services.

[0008] In order to achieve the above object, the technical solution specifically adopted by the present invention is as follows:

[0009] A method for realizing perception based on SSB signal in 5G NR comprises the following steps:

[0010] S1 and RRU periodically transmit SSB scanning signals to search for new users, help new users access and synchronize the time and frequency of other users. At the same time, users and targets in the environment reflect the transmitted signals back;

[0011] S2, the RRU receives the signal reflected in step S1, strips the data from the reflected signal, writes it into a compressed sensing form, and converts it into a delayed grid model;

[0012] S3, performing a unitary transformation on the signal processed in step S2, and then estimating the multipaths with different delays by using the UAMP-SBL algorithm;

[0013] S4, for the case where multiple multipaths have the same delay, write the multipath estimation result obtained in step S3 into a compressed sensing form;

[0014] S5. For the signal form obtained in step S4, use the OGSBI algorithm to separate and estimate the multipaths at different angles;

[0015] S6. According to the estimated angle and the transmission angles of different RRUs during transmission, it is determined to which RRU each set of parameters belongs.

[0016] Furthermore, in step S1, assuming that there are a total of G SSBs in a subframe, the signal on the nth subcarrier of the kth (k=1, 2, 3) OFDM symbol in the gth (g=1, ..., G) SSB can be expressed as:

[0017] x q,n,t =w q,t s n,t ,

[0018] Among them, s n,t is the modulated OFDM signal transmitted on the nth subcarrier in t=4g+k symbols, assuming that its energy is a constant of 1; w q,t is the direction θ transmitted by the qth RRU q,g The beamforming vector in the direction.

[0019] Furthermore, in step S2, the modulated signal in the received signal is first removed to obtain a processed received signal r n,t :

[0020]

[0021] Among them, r n,t is the received signal y on the n subcarriers of the tth symbol n,t After the modulation signal is removed, b q,l is the channel coefficient, f D,q,l is the Doppler frequency shift, Δf is the subcarrier spacing, θ q,l is the launch angle or arrival angle, is the processed noise;

[0022] Then, r on all subcarriers n,t Arrange them to get the matrix R formed by stacking the signals on all subcarriers processed at the same time in rows t :

[0023] R t =[r 1,t ,...,r N,t ] T =DB t +Z t ,

[0024] in, Each column in is the delay τ of the lth multipath of the qth RRU q,l Decision, there are Q RRUs in total, and the qth RRU has L q multipath, and

[0025] B t is composed of all paths The matrix formed is:

[0026] Indicates the Lth of the Qth RRU Q Multipath;

[0027] Let the total number of multipaths Noise after processing The noise matrix formed by stacking And b q,l,t τ in q,l Δf is quantified as N d is the number of quantized grids, l′ is the grid position of the lth multipath in the dictionary, then D can be written in the form of increasing delay in the grid:

[0028]

[0029] in, is the quantized τ q,l Δf, is c(τ q,l Δf) in quantitative form, N p (L<<N p <N d ) is the number of grids used, and then we get the following formula:

[0030] R t =D′PB t +Z t ,

[0031] Among them, PB t Estimated by sparse signal recovery methods, the matrix P is an N matrix with only L non-zero rows. p ×L-dimensional matrix, each non-zero row has only one non-zero value 1, which is used to convert B t Each multipath in B is mapped to the corresponding delay in the grid; t contains all the parameters that need to be estimated, B t Each row of corresponds to each delay in D′.

[0032] Furthermore, in step S3, the three OFDM signals in the g-th SSB are first arranged to obtain:

[0033] R g =[R t′ ,R t′+1 ,R t′+2 ]=D′PB g +N g ,

[0034] Where, t′=4g+1,B g =[B t′ ,B t′+1 ,B t′+2 ],N g =[Z t′ ,Z t′+1 ,Z t′+2 ], B g 、N g They are respectively the B corresponding to the three signals in the g-th SSB t and Z t Arranged and PB estimated using the UAMP-SBL algorithm g ;

[0035] Then, all PB g Arrange it up:

[0036] B=[PB1,PB2,...,PB G ],

[0037] Then, filter out L rows that are greater than the threshold and get:

[0038] B′=P′[B1,...,B G ],

[0039] Among them, B′ is the matrix reconstructed after filtering out the rows in B whose row energy is greater than the threshold, and P′ is the matrix obtained after removing the all-0 rows in P;

[0040] Finally, let the row index in B' be r, and the row index corresponding to r in B be r'. Then, the estimated value of the delay is It can be expressed as:

[0041]

[0042] Furthermore, in step S4, row index r is used to replace q and l, and the three signals in one SSB are taken as a group, and there are in Each row in B′ represents the multipath with the corresponding delay;

[0043] Assuming that a delay may have multiple multipaths, then:

[0044]

[0045] in, The i in represents the i-th path with the same delay, and I r It refers to the total number of paths corresponding to the rth delay;

[0046] Let the rth row, the gth part Transpose, let So there is

[0047]

[0048] in, is the matrix consisting of the angles in the summation, S r,g =[s r,t′ ,s r,t′+1 ,s r,t′+2 ] is the matrix of the remaining terms in the summation, and s r,t is an I r A vector of dimension , where each element V of all G parts r,g Put together, there is V r =[V r,1 ,V r,2 ,...,V r,G ],S r =[S r,1 ,S r,2 ,...,S r,G ],but:

[0049] V r =A r S r ,

[0050] Using the off-grid method to estimate the parameters, let as a grid of angles, where N A Represents the number of grids used, and then Representative The nearest grid point, approximated using a Taylor expansion first-order linear approximation but:

[0051]

[0052] in, is the array response vector of the uniform linear array, is the Taylor expansion point, represent The derivative of

[0053] definition Then Φ(β)=A D +Ωdiag(β), where

[0054]

[0055] Ignoring the approximation error, we get:

[0056] V r =Φ(β)X r ;

[0057] Among them, X r =PS r , the P matrix is ​​used to transform S r A matrix that maps to the corresponding delay grid.

[0058] Further, in step S5, first, based on V r =Φ(β)X r , the angle is calculated by OGSBI algorithm:

[0059]

[0060] in, is the estimated angle of the i-th multipath corresponding to the r-th delay, is the grid in the angle grid dictionary that is closest to this angle, and is the estimated angle offset relative to the fixed grid. Then, reconstruct A r matrix:

[0061] Using the least squares method, we can obtain The estimated value S r :

[0062]

[0063] S r Written in the following form:

[0064]

[0065] in, It corresponds to For the SSB part of the angle, select the SSB with the largest energy;

[0066] Assume that the SSB with the largest energy of the i-th multipath corresponding to the r-th delay is denoted by g. r,i , then g r,i It can be obtained from the following formula:

[0067]

[0068] because The estimated value of Doppler frequency can be obtained

[0069]

[0070] Finally, due to You can get:

[0071]

[0072] Furthermore, in step S6, it is determined and estimated which RRU each set of parameters belongs to based on the following formula:

[0073]

[0074]

[0075] The present invention has the following characteristics and beneficial effects:

[0076] The present invention proposes a new method for realizing perception function based on SSB signal in PMN, which estimates parameters by using UAMP-SBL and OG-SBL algorithms, and solves the problem of parameter matching through the estimated angles, which can significantly improve the perception accuracy and reduce the computational complexity, and has higher robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0078] Figure 1 This is a flow chart of a method for realizing perception based on SSB signals in 5G NR according to an embodiment of the present invention;

[0079] Figure 2 The estimated angle of arrival and distance and the angle of launch and distance for the present invention and the prior art;

[0080] In the figure: (a) arrival angle and distance estimated by the present invention; (b) arrival angle and distance estimated by existing inventions; (c) emission angle and distance estimated by the present invention; (d) emission angle and distance estimated by existing inventions.

[0081] Figure 3 Doppler and distance estimated for the present invention and prior art;

[0082] In the figure: (a) Doppler and distance estimated by the present invention; (b) Doppler and distance estimated by existing inventions.

[0083] Figure 4 This is a graph showing the estimated hit rate results for the present invention and the prior art.

[0084] Figure 5 This is a comparison chart of the accuracy of the estimation of the present invention and the existing invention, and the comparison indicator is the root mean square error. DETAILED DESCRIPTION

[0085] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0086] The embodiment of the present invention provides a new method for realizing the sensing function based on SSB signals in PMN, which senses the environment around multiple remote radio units (RRUs) based on the existing synchronization signal blocks (SSBs) in 5G NR, combining downlink active sensing and passive sensing. Active sensing means that RRU can receive the signal reflected back after it is transmitted, while passive sensing means that RRU can receive the signal reflected back after being transmitted by other RRUs. The specific flow chart is as follows: Figure 1 As shown, the following steps are included:

[0087] S1 and RRU periodically transmit SSB scanning signals, and users and targets in the environment reflect the transmitted signals back;

[0088] S2, the RRU receives the signal reflected in step S1, strips the data from the reflected signal, writes it in the form of a compressed sensing problem, and converts it into a delay in a grid model;

[0089] S3, performing a unitary transformation on the signal processed in step S2, and then estimating the multipaths with different delays by using the UAMP-SBL algorithm;

[0090] S4, for the case where multiple multipaths have the same delay, the estimation result in step S3 is further written into a compressed sensing form;

[0091] S5. For the signal form in step S4, use the OGSBI algorithm to separate and estimate the multipaths at different angles;

[0092] S6. According to the estimated angle and the transmission angles of different RRUs during transmission, it is determined to which RRU each set of parameters belongs.

[0093] More specifically, we consider a scenario where we consider implementing sensing in a PMN based on a cloud radio access network (CRAN). Typically, the network consists of a CRAN central and multiple RRUs. The CRAN central includes a baseband unit (BBU) pool and a sensor processing unit, which are used to process the cellular communication and wireless sensing signals collected from the RRUs, respectively. The densely distributed RRUs are usually connected to the CRAN central via wired cables and synchronized on the clock to ensure efficient and coordinated processing. It is assumed that the cooperating RRUs use multi-user multiple input multiple output (MIMO) and orthogonal frequency division multiple access (OFDMA) technologies to serve user equipment (UE), which are located within the coverage area of ​​the RRUs. This enables the PMN to leverage the collective power of multiple RRUs for accurate sensing.

[0094] Here, we assume a PMN with Q RRUs, each using a uniform linear array (ULA) with M antenna elements with half-wavelength spacing. During SSB communication, each RRU transmits an SSB burst set consisting of multiple SSBs in different beamforming directions. Let N be the total number of subcarriers and B be the total bandwidth. The subcarrier spacing is Δf = B / N, and the duration of an OFDM symbol is T. s =N / B+T p , where T p is the duration of the cyclic prefix.

[0095] Assuming the existence of a plane wavefront in signal propagation, the array response vector a(M,θ) of a uniform linear array (ULA) is given by:

[0096] a(M,θ)=[1,e jπsinθ ,...,e jπ(M-1)sinθ ] T , (1)

[0097] Where j is a complex number, M is the number of antennas, and T represents transposition. θ can be the angle of transmission (AoD) or the angle of arrival (AoA). Let multipath AoD and multipath AoA be denoted by θ l and φ l , (l=1,...,L), where L is the number of multipaths. T transmit antennas and M R receiving antennas, M R ×MT The time domain baseband channel matrix At the time When can be expressed as:

[0098]

[0099] in, is the impulse response, a(M R ,φ) is the number of receiving antennas M R An array response vector, a T (M T ,θ l ) is the transmitting antenna M T The array response vector. b l It is l The complex amplitude of the multipath, taking into account signal attenuation and initial phase difference, τ l is the propagation delay, f D,l is the relevant Doppler frequency. The sensor parameter to be estimated {τ l ,f D,l ,φ l ,θ l ,b l} can be used to determine the location and motion state of the target and UE. It is assumed that the channel is static within a few milliseconds, that is, these sensing parameters remain unchanged during this period. The Doppler phase of a sampling point in one OFDM symbol is approximately constant.

[0100] We process the received signal in the frequency domain. For the t-th OFDM symbol, from (2), the frequency domain channel matrix H corresponding to the n-th subcarrier is n,t It can be expressed as:

[0101]

[0102] The Doppler phase of a sampling point in one OFDM symbol is approximately constant. For downlink sensing, each RRU receives the reflected downlink signal from itself and other Q-1 RRUs, so the signal y received at the nth subcarrier and tth OFDM symbol is n,t It can be expressed as:

[0103]

[0104] Among them, L q is the multipath number of the qth RRU, x q,n,t and z n,t are the transmitted signal vector and noise vector of the qth RRU at the nth subcarrier and the tth symbol respectively.

[0105] In step S1: Due to the broadcast characteristics and periodic transmission of SSB in 5G NR, it has a great advantage for perception. Here, we use SSB signal. Assuming that there are G SSBs in a subframe, the signal on the nth subcarrier of the kth (k=1,2,3) OFDM symbol in the gth (g=1,...,G) SSB can be expressed as:

[0106] x q,n,t =w q,t s n,t , (5)

[0107] Among them, s n,t is the modulated OFDM signal transmitted on the nth subcarrier in t=4g+k symbols, and we assume that its energy is a constant of 1. q,t is the direction θ transmitted by the qth RRU q,g In order to estimate the AoD, we let

[0108]

[0109] Among them, w m is the weight of the mth antenna (1≤m≤M),

[0110] In step S2: we remove the modulated signal from the received signal and obtain the processed received signal r n,t :

[0111]

[0112] Among them, r n,t is the received signal y on the n subcarriers of the tth symbol n,t The signal after the modulation signal is removed, where: b q,l is the channel coefficient, f D,q,l is the Doppler frequency shift, Δf is the subcarrier spacing, θ q,l is the launch angle or arrival angle. is the processed noise; then, we can stack the processed signals on all subcarriers at the same time in rows to form the matrix R t :

[0113] R t =[r 1,t ,...,r N,t ] T =DB t +Z t , (8)

[0114] in,

[0115]

[0116] Each column in D is the delay τ of the lth multipath of the qth RRU. q,l Decision, there are Q RRUs in total, and the qth RRU has L q multipath, and

[0117]

[0118] B t is composed of all paths The matrix formed, Indicates the Lth RRU of the Qth RRU at the tth time Q multipaths. We let the number of all multipaths Noise after processing The noise matrix formed by stacking At the same time, b q,l,t τ in q,l Δf is quantified as N d is the number of quantized grids, l′ is the grid position of the lth multipath in the dictionary, then D can be written in the form of increasing delay in the grid:

[0119]

[0120] in, This corresponds to the result of the minimum delay quantization. is the quantized τ q,l Δf, is c(τ q,l Δf), that is, τ q,l After quantization, the corresponding l'th column in D' is N p (L<<N p <N d ) is the number of grids used, then we can rewrite (8) as:

[0121] R t =D′PB t +Z t , (12)

[0122] Here, the matrix P is an N matrix with only L non-zero rows. p ×L-dimensional matrix, each non-zero row has only one non-zero value 1, which is used to convert B t Each multipath in B is mapped to a corresponding delay in the grid. t contains all the parameters we need to estimate.t Each row of corresponds to each delay in D′. Due to the sparseness of multipath in the delay domain, PB t Only L of them have values. So we can regard (12) as a coefficient signal recovery problem. By using the sparse signal recovery method, we can recover PB t It can be estimated, and then the multipath parameters inside can be estimated.

[0123] In step S3: For model (12), we can arrange the three OFDM signals in the g-th SSB to make full use of them, and we have:

[0124] R g =[R t′ ,R t′+1 ,R t′+2 ]=D′PB g +N g , (13)

[0125] where t′=4g+1,B g =[B t′ ,B t′+1 ,B t′+2 ],N g =[Z t′ ,Z t′+1 ,Z t′+2 ], B g and N g There are three signals in the g-th SSB corresponding to B t and Z t Obviously, (13) can be regarded as a sparse signal recovery problem, and we can use the UAMP-SBL algorithm to estimate PB g . Note that PB g However, the gth SSB is only aimed at a certain direction, which may not hit the target, so the echo may be too small to be filtered out. g Arrange

[0126] B=[PB1,PB2,...,PB G ], (14)

[0127] In this way, we can ensure that every place is scanned. Then, we filter out the L rows that are greater than the threshold, reconstruct the matrix, and get B′:

[0128] B′=P′[B1,...,B G ], (15)

[0129] Where P′ is the matrix obtained by removing all zero rows in P.

[0130] We let the row index in B' be r, and the row index corresponding to r in B be r'. Then, the estimated value of the delay is It can be expressed as:

[0131]

[0132] In step S4: we will consider the more general case where multiple multipaths have the same delay to estimate the parameters. Since we do not know b q,l In the example, q and l have specific values, so we use row index r to replace q and l, and take the three signals in an SSB as a group, in Each row in B′ represents the multipath of the corresponding delay. Assuming that a delay may have multiple multipaths, then

[0133]

[0134] in, The i in represents the i-th path with the same delay, and I r It refers to the total number of paths corresponding to the rth delay; represents the multipath corresponding to the rth delay at time t', and an SSB is composed of three consecutive OFDM signals. We write the three signals together to form by The first block in For example, it can be expressed as:

[0135]

[0136] in, and is the parameter of the i-th path with the same delay corresponding to the r-th delay.

[0137] We let the rth row, the gth part Transpose, let So there is

[0138]

[0139] in, is the matrix S consisting of the angles in the summation r,g =[s r,t′ ,s r,t′+1 ,s r,t′+2 ] is the matrix of the remaining terms in the summation, and s r,t is an I r A vector of dimension , where each element We will all G parts of Vr,g Put together, there is V r =[V r,1 ,V r,2 ,...,V r,G ],S r =[S r,1 ,S r,2 ,...,S r,G ], so we have

[0140] V r =A r S r , (20)

[0141] This can be further constructed as a sparse signal recovery problem.

[0142] Since the angle is not like the delay, the error will not be large even on the grid. We use the off-grid method to estimate the parameters. We let as a grid of angles, where N A Represents the number of grids used, and then Representative The nearest grid point, we use a linear approximation to approximate have

[0143]

[0144] in, represent The derivative of . We define besides Then we have Φ(β)=A D +Ωdiag(β), where

[0145]

[0146] Ignoring the approximate error, we can easily obtain:

[0147] V r =Φ(β)X r , (twenty three)

[0148] Where X r =PS r , the P matrix has the same function as before, which is to convert S r A matrix that maps to the corresponding delay grid.

[0149] In step S5: We use the OGSBI algorithm for (23) and it is not difficult to get the angle:

[0150]

[0151] After obtaining the angle, we want to further estimate the angle contained in S r , so we reconstruct A r matrix:

[0152]

[0153] Then we use the least squares method to obtain:

[0154]

[0155] We will S r Written in this form:

[0156]

[0157] in, It corresponds to For the SSB part of the angle, we choose the SSB with the largest energy, g r,i Make an estimate, that is:

[0158]

[0159] because We can get Doppler

[0160]

[0161] Then, according to formula (6), we can get:

[0162]

[0163] In step S6: Now we have all the parameters However, we do not know which RRU's multipath these parameters come from, so we can start from the angle range and get the i-th multipath of the r-th delay from

[0164]

[0165] In addition, we can also estimate the energy of the multipath:

[0166]

[0167] in, represents the energy of the i-th multipath corresponding to the r-th delay, It is to In addition to the results Remove the items of to get

[0168] In this section, the performance of the proposed sensing method is analyzed and compared with the existing BSBL methods. We consider a PMN with two remote radio units (RRUs), each equipped with 4 antenna elements, which are synchronized and coordinated to send synchronization signal block (SSB) groups in different directions and strategically positioned to ensure that they are within each other's signal coverage. The system operates at a carrier frequency of 2.35 GHz with a signal bandwidth of 100 MHz. For downlink sensing, we use SSB as the sensing signal, which is designed to occupy 240 subcarriers and 12 symbols in the time-frequency domain. We assume the radar cross section (RCS) of the target is 1 square meter in the simulation. The propagation paths are randomly generated in the clusters, following a complex Gaussian distribution, and a path loss model with a path loss factor of 4 is adopted. The transmit power of the RRU is set to 35 dBm. We assume that the received signal is contaminated by additive white Gaussian noise (AWGN) with a thermal noise power of N0 = -174 dBm / Hz. Therefore, the total thermal noise at the receiver is -174+10log(108)=-94dBm. The directional span, distance, and moving speed of the multipath are assumed to be uniformly distributed in the range of [-60,60] degrees, [15,150] meters, and [-30,30] meters / second. The delay is quantized on a grid with an interval of 10 nanoseconds, corresponding to a distance of 3 meters. Based on the path loss factor and the multipath propagation distance, we can see that the received signal-to-noise ratio (SNR) used to estimate the sensing parameters for each path may be 5dB, assuming that the sensing parameters are constant within a channel coherence period.

[0169] Figure 2 and Figure 3 The results of ten estimations of AoA-distance, Doppler-distance and AoD-distance are shown. The figure shows a scatter plot of the estimated values ​​and the true values ​​of the perception parameters, where the circle symbol, plus symbol and multiplication symbol represent the actual value, the value estimated by the present invention and the value estimated by the BSBL method respectively. Different colors are used to distinguish different clusters in the environment. We assume that the number of multipaths of each cluster is randomly distributed between [1,4], simulating the signal reflected / scattered by the target, and the delay of each path is different. It can be clearly seen that the present invention is superior to the BSBL method in terms of the accuracy and stability of the estimation results. This is because the present invention utilizes the robustness of UAMP, avoids the divergence of AMP, and implements SBL through structured variational inference, and utilizes common support. In addition, the proposed method also has a lower UAMP complexity than the BSBL method. The results show that our method is superior to the BSBL method in terms of both recovery performance and computational efficiency.

[0170] Next, we evaluated the performance of object detection. Figure 4An in-depth statistical comparison of hit rates is presented for signal-to-noise ratios (SNRs) ranging from -15dB to 5dB, where the hit rate is defined as the ratio of the number of correctly detected targets to the number of actual targets. The comparison results show that the hit rate of the present invention is always higher than that of the BSBL method regardless of the SNR value. Figure 5 The estimation performance of other parameters of the two inventions under different SNR conditions is demonstrated. The results show that the parameter estimation accuracy of the present invention is better than that of the BSBL method under all SNR levels.

[0171] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for realizing perception based on SSB signal in 5G NR, characterized in that: The steps include: S1 and RRU periodically transmit SSB scanning signals to search for new users, help new users access and synchronize the time and frequency of other users. At the same time, users and targets in the environment reflect the transmitted signals back; S2, the RRU receives the signal reflected in step S1, strips the data in the reflected signal, writes it into a compressed sensing form, and converts it into a delayed grid model; S3, performing a unitary transformation on the signal processed in step S2, and then estimating the multipaths with different delays by using the UAMP-SBL algorithm; S4. For the case where multiple multipaths have the same delay, the multipath estimation result obtained in step S3 is written in a compressed sensing form; S5. For the signal form obtained in step S4, use the OGSBI algorithm to separate and estimate the multipaths at different angles; S6. According to the estimated angle and the transmission angles of different RRUs during transmission, it is determined to which RRU each set of parameters belongs.

2. A method for realizing perception based on SSB signals in 5G NR according to claim 1, characterized in that: In step S1, assuming that there are a total of G SSBs in a subframe, the signal on the nth subcarrier of the kth (k=1,2,3) OFDM symbol in the gth (g=1,...,G) SSB can be expressed as: x q,n,t =w q,t s n,t ; Among them, s n,t is the modulated OFDM signal transmitted on the nth subcarrier in t=4g+k symbols, assuming that its energy is a constant of 1; w q,t is the direction θ transmitted by the qth RRU q,g The beamforming vector in the direction.

3. The method for realizing perception based on SSB signal in 5G NR according to claim 1, characterized in that: In step S2, the modulated signal in the received signal is first removed to obtain a processed received signal: Among them, r n,t is the received signal y on the n subcarriers of the tth symbol n,t After the modulation signal is removed, b q,l is the channel coefficient, f D,q,l is the Doppler frequency shift, Δf is the subcarrier spacing, θ q,l is the launch angle or arrival angle, is the processed noise; Then, r on all subcarriers n,t Arrange them to get the matrix R formed by stacking the signals on all subcarriers processed at the same time in rows t : R t =[r 1,t ,...,r N,t ] T =DB t +Z t , in, Each column in is the delay τ of the lth multipath of the qth RRU q,l Decision, there are Q RRUs in total, and the qth RRU has L q multipath, and B t is composed of all paths The matrix formed is: Indicates the Lth of the Qth RRU Q Multipath; Let the total number of multipaths Noise after processing The noise matrix formed by stacking And b q,l,t τ in q,l Δf is quantified as Among them, N d is the quantization grid number, l′ is the grid position of the lth multipath in the dictionary; Then D can be written in delayed-incremental grid form: in, is the quantized τ q,l Δf, is c(τ q,l Δf) in quantitative form, N p (L<<N p <N d ) is the number of grids used, and then we get the following formula: R t =D′PB t +Z t , Among them, PB t Estimated by sparse signal recovery methods, the matrix P is an N matrix with only L non-zero rows. p ×L-dimensional matrix, each non-zero row has only one non-zero value 1, which is used to convert B t Each multipath in B is mapped to the corresponding delay in the grid; t contains all the parameters that need to be estimated, B t Each row of corresponds to each delay in D′.

4. The method for realizing perception based on SSB signal in 5G NR according to claim 1, characterized in that: In step S3, firstly, the three OFDM signals in the g-th SSB are arranged to obtain: R g =[R t′ ,R t′+1 ,R t′+2 ]=D′PB g +N g , Where t′=4g+1,B g =[B t′ ,B t′+1 ,B t′+2 ],N g =[Z t′ ,Z t′+1 ,Z t′+2 ], B g 、N g They are respectively the B corresponding to the three signals in the g-th SSB t and Z t Arranged and PB estimated using the UAMP-SBL algorithm g ; Then, all PB g Arrange it up: B=[PB1,PB2,...,PB G ], Then, select L rows greater than the threshold and reconstruct the matrix B′: B′=P′[B1,...,B G ], Where P′ is the matrix obtained by removing all zero rows in P; Finally, let the row index in B' be r, and the row index corresponding to r in B be r'. Then, the estimated value of the delay is It can be expressed as:

5. The method for realizing perception based on SSB signal in 5G NR according to claim 1, characterized in that: In step S4, row index r is used to replace q and l, and the three signals in one SSB are taken as a group. in ,Each row in B′ represents the multipath with corresponding delay; Assuming that a delay may have multiple multipaths, we can get: in, The i in represents the i-th path with the same delay, and I r It refers to the total number of paths corresponding to the rth delay; Let the rth row, the gth part Transpose, let Then we can get: in, is the matrix consisting of the angles in the summation, S r,g =[s r,t′ ,s r,t′+1 ,s r,t′+2 ] is the matrix of the remaining terms in the summation, and s r,t is an I r A vector of dimension , where each element V of all G parts r,g Put together, there is V r =[V r,1 ,V r,2 ,...,V r,G ],S r =[S r,1 ,S r,2 ,...,S r,G ],but: Using the off-grid method to estimate the parameters, let as a grid of angles, where N A Represents the number of grids used, and then Representative The nearest grid point, approximated using a Taylor expansion first-order linear approximation but: in, is the array response vector of the uniform linear array, is the Taylor expansion point, represent The derivative of definition Then Φ(β)=A D +Ωdiag(β), where: Ignoring the approximation error, we get: V r =Φ(β)X r ; Among them, X r =PS r , the P matrix is ​​used to transform S r A matrix that maps to the corresponding delay grid.

6. A method for realizing perception based on SSB signals in 5G NR according to claim 1, characterized in that: In step S5, first, based on V r =Φ(β)X r , the angle is calculated by OGSBI algorithm: in, is the estimated angle of the i-th multipath corresponding to the r-th delay, is the grid in the angle grid dictionary that is closest to this angle, and is the offset of the estimated angle relative to the fixed grid; Then, reconstruct A r matrix: Using the least squares method, we can obtain The estimated value S r : S r Written in the following form: in, It corresponds to For the SSB part of the angle, select the SSB with the largest energy; Assume that the SSB with the largest energy of the i-th multipath corresponding to the r-th delay is denoted by g. r,i , then g r,i It can be obtained from the following formula: because The estimated value of Doppler frequency can be obtained Finally, due to You can get:

7. The method for realizing perception based on SSB signal in 5G NR according to claim 1, characterized in that: In step S6, it is determined and estimated which RRU each set of parameters belongs to based on the following formula:

Images