Uplink sensing iterative optimization method based on cellular-free architecture

Abstract

The invention discloses an uplink sensing iterative optimization method based on a cellular-free system, which is used for solving the problems that the sensing precision depends on the communication demodulation quality and the communication performance is limited by the sensing result in an ISAC system, and is suitable for future 6G scenes of large-scale terminal connection and high-precision sensing cooperation. The method comprises the following steps: constructing a multi-signal collaborative sensing module, fusing pilot frequency initial sensing and data signal enhanced sensing, and extracting target parameters; constructing a dynamic channel reconstruction module, and mapping a sensing result into channel state information required by communication; and constructing a robust data demodulation module, recovering communication data by using channel information, and feeding back a demodulation error to correct a sensing result. The above modules form a closed-loop feedback mechanism, and the sensing precision and the communication reliability are gradually improved in the iteration process. Compared with the prior art, the method can effectively reduce the bit error rate and improve the sensing positioning precision, has the advantages of low cost, high accuracy and the like, and has wide practical application value and engineering popularization prospect.

Description

Technical Field

[0001] This invention relates to the field of distributed MIMO system technology, and in particular to an uplink sensing iterative optimization method based on a cellular-free architecture. Background Technology

[0002] In future 6G communication systems, networks will face multiple demands, including massive terminal access, high-precision sensing, ultra-low latency, and high reliability. Integrated Communication and Sensing (ISAC), as a key technology supporting intelligent wireless services, has become a research hotspot. Traditional cellular architectures are limited by issues such as cell division, edge interference, and resource silos, making it difficult to meet the future needs of dense coverage and wide-area collaborative sensing. In contrast, cell-free architectures, through large-scale distributed access points, collaboratively serve all users, offering advantages such as no cell handover, high coverage, and low latency, demonstrating great potential in building integrated systems that combine high-precision sensing and high-reliability communication. Therefore, integrated communication and sensing solutions based on cell-free systems have become an important direction for building next-generation intelligent wireless networks.

[0003] However, in cellular-free systems, communication and sensing tasks typically share user uplink resources. Communication signals are multiplexed for sensing, and sensing results are used for channel estimation, creating a natural coupling between the two. Sensing performance depends on reliable demodulation of communication signals, which in turn heavily relies on the accurate acquisition of channel state information, which must be inferred from the sensing module. Therefore, traditional methods of modeling and processing communication and sensing separately are no longer sufficient to meet the needs of system collaborative optimization. Furthermore, in real-world scenarios with limited spectrum and pilot resources, pilot conflicts and signal interference among multiple uplink users significantly impact channel estimation accuracy, thereby affecting communication performance and the reliability of target sensing. Faced with dynamically changing environments and complex user distributions, relying solely on single-sense sensing or fixed channel estimation methods is insufficient to meet the demands of high-precision sensor-sensing collaboration.

[0004] Existing ISAC-related research largely focuses on signal design or resource allocation, lacking in-depth joint modeling and feedback mechanism design for the coupling relationship between communication and sensing. While some methods consider sensing-assisted communication or communication-enhanced sensing, they are often unidirectional optimizations, unable to achieve closed-loop improvement in multiple iterations, and prone to performance bottlenecks in dynamic environments. Therefore, there is an urgent need for an iterative method that can establish an alternating optimization mechanism between communication demodulation, channel reconstruction, and sensing estimation, achieving system-level closed-loop synergy of sensing-driven communication enhancement and communication feedback-based sensing refinement in multiple processing rounds. This would effectively overcome the technical bottlenecks of high communication error rates and insufficient sensing accuracy in traditional ISAC systems under complex environments. Summary of the Invention

[0005] In view of this, this invention proposes an uplink sensing iterative optimization method based on a cellular-free architecture. By establishing a joint modeling and closed-loop feedback mechanism among communication demodulation, target perception, and channel reconstruction, it achieves a synergistic improvement in communication and sensing performance. This method utilizes user uplink signals to simultaneously complete data transmission and multi-target perception. It employs three major modules—multi-signal collaborative perception, dynamic channel reconstruction, and robust data demodulation—that alternately and iteratively update, forming an iterative optimization process where communication errors drive the improvement of sensing accuracy, and sensing results assist in channel reconstruction. This improves both communication reliability and sensing accuracy, making it suitable for next-generation wireless communication systems with large-scale access and high-precision sensing requirements.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] An uplink sensing iterative optimization method based on a cellular-free system is disclosed. In the distributed MIMO system, there are M access points (APs), each equipped with S antennas, and K single-antenna users. Users and targets are randomly distributed within the coverage area. The communication channel includes one Loss-of-Sight (LoS) path and T Non-LoS (NLoS) paths. Each NLoS path consists of a two-segment path: user-target-access point. The method specifically includes the following steps:

[0008] Step S1: Establish an uplink non-cellular integrated communication and sensing (CF-ISAC) system model. The user sends pilot signals for initial channel estimation and sends data to construct the received signal model, thereby obtaining the basic observation data required for communication and sensing.

[0009] Step S2: Using the initial channel estimation and received data signal, three tasks are performed in each iteration: First, the data symbols are recovered using the improved D-AMP algorithm; then, the angle and distance of the target are estimated using the 2D-MUSIC algorithm with the recovered data and pilot estimation as input; finally, the sensing parameters are used to reconstruct the current communication channel using the geometric channel model.

[0010] Step S3: The data demodulation residual feedback is used for parameter search and adjustment in the sensing module, so that the communication error optimizes the sensing accuracy in reverse. The three modules drive and correct each other in each round, and continue to iterate until the communication bit error rate and the changes in sensing parameters meet the preset convergence conditions, and finally output the optimization result.

[0011] Furthermore, step S1 specifically includes:

[0012] Step S101: Establish an uplink cellless integrated communication and sensing (CF-ISAC) system model. Under the nth subcarrier, the channel between the mth AP and the kth user is...

[0013]

[0014] Where, λ m,t,k q represents the large-scale fading factor. t,m q k,t These are the direction vectors for the target, AP, and user, respectively, d m,k,t Here, Δf represents the distance information, Δf represents the subcarrier spacing, c represents the speed of light, and T represents the number of scatterers in the channel.

[0015] Step S102: First, user k sends an orthogonal pilot signal of length . The m-th AP receives the pilot signal as follows: The initial channel estimation is performed using the LS method based on the pilot signal.

[0016]

[0017] Among them, P p It is the pilot transmission power, N m It is additive Gaussian noise; the initial channel estimation on N subcarriers is defined as... τ p It is the pilot length. It is the pilot sequence of the kth user.

[0018] Step S103: Similarly, user k sends data x. k Under the nth subcarrier, the data signal received by the mth AP is

[0019]

[0020] Among them, P d Here, z is the data transmission power, z is the additive Gaussian noise, and the received data matrix on N subcarriers is defined as follows:

[0021]

[0022] in, It is a matrix containing distance information. It is an angle information matrix, Z∈C MS×N It is a noise matrix.

[0023] Furthermore, step S2 specifically includes:

[0024] Step 201: The uplink CF-ISAC signal processing includes two core tasks: communication data decoding and target parameter estimation. Communication data decoding involves two aspects: first, recovering data symbols from the uplink communication signal; and second, extracting angle and distance information from the channel information based on pilot and data signals. The goal is to minimize the error between the reconstructed received signal and the actual received signal. Therefore, the ISAC signal processing problem can be expressed as follows:

[0025]

[0026] Step 202: First, based on the received data signal and the estimated channel, the improved D-AMP algorithm is used to recover the transmitted data symbols, mainly involving two core steps: noise update and signal update.

[0027]

[0028] Where j is the number of runs of the improved D-AMP algorithm. η1 is the weighting factor for incorporating historical iteration noise, and X... j This is the noise result from the previous iteration. It is a matrix containing distance information. It is the angle information matrix, and η2 is the weighting factor for fusing historical iteration signals. It is the noise update output of the k-th user in the j-th iteration.

[0029] Step 203: Pilot symbols and data symbols each occupy a portion of time-frequency resources. Since the CPU has known data signals, pilot symbols can be directly used for initial channel estimation to obtain... By recovering the data symbols through an improved D-AMP, and then removing the data symbols from the received information symbols using an element-wise division method, we can obtain... Therefore, the channel matrix used for target sensing can be represented as

[0030]

[0031] in For pilot signal receiving matrix, This represents the matrix obtained by removing data symbols from the received information symbols through element-wise division, where Q represents the total number of symbols.

[0032] The angle and distance are calculated separately based on the 2D-musc algorithm. Firstly, through... Construct the angle spectrum function,

[0033]

[0034] Where a(θ) is the angular direction vector, and E0 is... The noise subspace of the covariance matrix. F -1 The angle corresponding to the t-th maximum value of (θ) is the angle of the t-th target. Next, a normalized filter vector w is designed for each angle. t Thus, the channel matrix at this angle is constructed. pass Construct the distance spectrum function,

[0035]

[0036] Where a(d) is the distance direction vector, U tN yes The noise subspace of the covariance matrix. The target distance at this angle is...

[0037] Step 204: By using the perception algorithm to obtain the location information of each target, including angle information and distance information, the relative distance and angle between the AP and the target, as well as between the target and the user, can be derived. Substituting these into the channel formula, the phase information and amplitude information of the channel can be recovered.

[0038] Furthermore, step S3 specifically includes:

[0039] Step S301: Input the received data signal, initial channel estimation, and number of iterations into the iterative optimization method, and initialize the data symbols and target estimation parameters;

[0040] Step S302: In the (l+1)th iteration, execute the three modules sequentially.

[0041] Sub-step 1: In the multi-signal collaborative sensing module, recover the data symbols obtained in the l-th iteration. The channel matrix for this iteration is constructed based on the initial channel estimation, and the 2D-music algorithm is used to estimate the target's orientation. and distance parameters Historical perception results We obtain a weighted fusion with the current estimation results.

[0042] Sub-step 2: Utilize the target parameter estimation results in the dynamic channel reconstruction module. The location relationship between the user and the AP is used to reconstruct the path components in the geometric channel model, and the reconstructed channel is calculated. Furthermore, the reconstruction results are weighted historically to obtain... η4 is the weighting factor for reconstructing the channel by fusing historical iterations.

[0043] Sub-step 3: In the robust data demodulation module, based on the currently obtained channel estimation An improved D-AMP algorithm is adopted to obtain the demodulated symbol X through noise update and signal update. l+1 and demodulation residuals Furthermore, the demodulation residual is fed back to the perception module in the next iteration to dynamically adjust the search step size in 2D-music.

[0044] Step S303: Terminate the iteration if the following conditions are met:

[0045] BER l+1-BER l ||≤∈ comm ,‖A l+1 -A l ||≤∈ sens

[0046] Where ∈ comm and ∈ sens These are the performance thresholds for communication and sensing, respectively. If they are not met, repeat step S302.

[0047] The beneficial effects of this invention are as follows: By constructing an iterative optimization mechanism between communication demodulation, target perception, and channel reconstruction, this invention achieves a synergistic improvement in communication and perception functions, effectively solving the problems of high communication error rate and low perception accuracy in traditional methods. In complex environments with limited resources and dynamically changing channels, it can significantly improve the accuracy of channel estimation and target positioning. It possesses low complexity, high robustness, and good scalability, making it suitable for large-scale access and high-precision perception scenarios, and has broad application prospects and engineering value. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the uplink sensing integrated architecture of the non-cellular distributed MIMO system provided in Example 1;

[0049] Figure 2 This is a flowchart illustrating the uplink sensing integrated iterative optimization method based on a non-cellular distributed MIMO system provided in Example 1.

[0050] Figure 3 To employ different signal-to-noise ratios, simulations were used to verify the changing trends of the bit error rate and distance estimation error of the proposed uplink inductive integrated iterative optimization method under different iteration numbers. Simulation graphs were used to verify the performance relationship between the number of iterations and the bit error rate and distance estimation error, respectively.

[0051] Figure 4 and Figure 5 This is a simulation graph comparing the bit error rate and distance estimation error performance of different sensing methods under different simulated signal-to-noise ratio conditions. It is used to verify the performance relationship between the uplink sensing integrated iterative optimization method, the LS-based channel estimation method, the single-layer iterative method, and the traditional non-iterative method (AMP estimation method). Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Example 1

[0054] See Figures 1-4 This embodiment provides an uplink synergistic sensing iterative optimization method based on a cellular-free architecture, specifically including:

[0055] like Figure 1 As shown, an uplink sensing iterative optimization method based on a cellular-free system is presented. In the distributed MIMO system, there are M access points (APs), each equipped with S antennas, and K single-antenna users. Users and targets are randomly distributed within the coverage area. The communication channel includes one Loss-of-Sight (LoS) path and T Non-LoS (NLoS) paths. Each NLoS path consists of a two-segment path from user to target to access point. The radar cross-section is 0.1, the user transmit power is 0.4 watts, the subcarrier width is 60 kHz, the data signal modulation scheme is 64QAM, the number of subcarriers is 128, and the cyclic prefix length is 12.

[0056] Based on the non-cellular distributed massive MIMO system provided above, and the constructed uplink received signal model, the uplink sensing iterative optimization method provided in this embodiment specifically includes the following steps:

[0057] Step S1: Establish an uplink non-cellular integrated communication and sensing (CF-ISAC) system model. The user sends pilot signals for initial channel estimation and sends data to construct the received signal model, thereby obtaining the basic observation data required for communication and sensing.

[0058] In this embodiment, step S1 specifically includes:

[0059] Step S11: Establish an uplink cellless integrated communication and sensing (CF-ISAC) system model. Under the nth subcarrier, the channel between the mth AP and the kth user is...

[0060]

[0061] Where, λ m,t,k q represents the large-scale fading factor. t,m q k,t These are the direction vectors for the target, AP, and user, respectively, d m,k,t Here, Δf represents the distance information, Δf represents the subcarrier spacing, and c represents the speed of light.

[0062] Step S12: First, user k sends an orthogonal pilot signal of length . The m-th AP receives the pilot signal as follows: The initial channel estimation is performed using the LS method based on the pilot signal.

[0063]

[0064] Among them, P p It is the pilot transmission power, N m It is additive Gaussian noise; the initial channel estimation on N subcarriers is defined as...

[0065] Step S13: Similarly, user k sends data x. k Under the nth subcarrier, the data signal received by the mth AP is

[0066]

[0067] Among them, P d Here, n is the data transmission power, and n is the additive Gaussian noise. The received data matrix on N subcarriers is defined as follows:

[0068]

[0069] in, It is a matrix containing distance information. It is an angle information matrix, Z∈C MS×N It is a noise matrix.

[0070] Step S2: Using the initial channel estimation and received data signal, three tasks are performed in each iteration: First, the data symbols are recovered using the improved D-AMP algorithm; then, the angle and distance of the target are estimated using the 2D-MUSIC algorithm with the recovered data and pilot estimation as input; finally, the sensing parameters are used to reconstruct the current communication channel using the geometric channel model.

[0071] In this embodiment, step S2 specifically includes:

[0072] Step 21: The uplink CF-ISAC signal processing includes two core tasks: communication data decoding and target parameter estimation. Communication data decoding involves two aspects: first, recovering data symbols from the uplink communication signal; and second, extracting angle and distance information from the channel information based on pilot and data signals. The goal is to minimize the error between the reconstructed received signal and the actual received signal. Therefore, the ISAC signal processing problem can be expressed as follows:

[0073]

[0074] Step 22: First, based on the received data signal and the estimated channel, the improved D-AMP algorithm is used to recover the transmitted data symbols. This mainly involves two core steps: noise update and signal update.

[0075]

[0076] Where j is the number of times the improved D-AMP algorithm is run.

[0077] Step 23: Pilot symbols and data symbols each occupy a portion of time-frequency resources. Since the CPU has known data signals, pilot symbols can be directly used for initial channel estimation to obtain... By recovering the data symbols through an improved D-AMP, and then removing the data symbols from the received information symbols using an element-wise division method, we can obtain... Therefore, the channel matrix used for target sensing can be represented as

[0078]

[0079] The angle and distance are calculated separately based on the 2D-musc algorithm. Firstly, through... Construct the angle spectrum function,

[0080]

[0081] Where a(θ) is the angular direction vector, and E0 is... The noise subspace of the covariance matrix. F -1 The angle corresponding to the t-th maximum value of (θ) is the angle of the t-th target. Next, a normalized filter vector w is designed for each angle. t Thus, the channel matrix at this angle is constructed. pass Construct the distance spectrum function,

[0082]

[0083] Where a(d) is the distance direction vector, U tN yes The noise subspace of the covariance matrix. The target distance at this angle is...

[0084] Step 24: By using the perception algorithm to obtain the location information of each target, including angle information and distance information, the relative distance and angle between the AP and the target, as well as between the target and the user, can be derived. Substituting these into the channel formula, the phase information and amplitude information of the channel can be recovered.

[0085] Step S3: The data demodulation residual feedback is used for parameter search and adjustment in the sensing module, so that the communication error optimizes the sensing accuracy in reverse. The three modules drive and correct each other in each round, and continue to iterate until the communication bit error rate and the changes in sensing parameters meet the preset convergence conditions, and finally output the optimization result.

[0086] The specific implementation steps are as follows:

[0087] Step S31: Input the received data signal, initial channel estimation, and number of iterations into the iterative optimization method, and initialize the data symbols and target estimation parameters;

[0088] Step S32: In the (l+1)th iteration, execute the three modules in sequence.

[0089] Sub-step 1: In the multi-signal collaborative sensing module, recover the data symbols obtained in the l-th iteration. The channel matrix for this iteration is constructed based on the initial channel estimation, and the 2D-music algorithm is used to estimate the target's orientation. and distance parameters Historical perception results We obtain a weighted fusion with the current estimation results.

[0090] Sub-step 2: Utilize the target parameter estimation results in the dynamic channel reconstruction module. The location relationship between the user and the AP is used to reconstruct the path components in the geometric channel model, and the reconstructed channel is calculated. Furthermore, the reconstruction results are weighted historically to obtain...

[0091] Sub-step 3: In the robust data demodulation module, based on the currently obtained channel estimation An improved D-AMP algorithm is adopted to obtain the demodulated symbol X through noise update and signal update. l+1 and demodulation residuals Furthermore, the demodulation residual is fed back to the perception module in the next iteration to dynamically adjust the search step size in 2D-music.

[0092] Step S33: Terminate the iteration if the following conditions are met:

[0093] BER l+1 -BER l ||≤∈ comm ,‖A l+1 -A l ||≤∈ sens

[0094] Where ∈ comm and ∈ sensThese are the performance thresholds for communication and sensing, respectively. If they are not met, repeat step S302.

[0095] The above demonstrates the entire process of iterative optimization of uplink sensing in a non-cellular system using the method provided in this embodiment.

[0096] Figure 3 The results clearly demonstrate the synergistic improvement in communication and sensing performance with increasing iteration count, fully reflecting the balance between performance enhancement and computational complexity. Specifically, the bit error rate (BER) and EME decrease rapidly at the beginning of the iteration, with the most significant performance improvements observed in the 3rd to 5th iterations, indicating that the information interaction and feedback mechanism plays a crucial role in the overall system performance during this stage. As iterations continue, the performance gain essentially converges by the 5th iteration. The results show that this method can achieve synergistic optimization of communication and sensing with a limited number of iterations, avoiding redundant computational burdens caused by excessive iterations. Furthermore, under different signal-to-noise ratio conditions, this method achieves convergence within 5 iterations through dynamically weighted historical information and adaptive feedback quantization of residuals, improving performance robustness.

[0097] Figure 4 and Figure 5 As shown, the LS channel estimation-based scheme is only applicable to pilot signal sensing and communication. Therefore, it achieves the worst performance as a benchmark scheme. The proposed iterative communication sensing optimization method achieves bidirectional gains in sensing and communication performance through external and internal iterative optimization. Compared with traditional non-iterative methods, this method continuously corrects sensing parameters through dynamic communication feedback, significantly reducing target localization errors. Simultaneously, it optimizes channel estimation and data demodulation using sensing results, significantly improving communication reliability. Compared with single-layer iterative methods, this method further introduces quantized residual feedback to adaptively adjust the sensing resolution and computational resource allocation, while balancing accuracy and efficiency in dense multipath environments. Furthermore, internal iteration accelerates the convergence speed of data detection, significantly reduces bit error rate fluctuations, and enhances transmission stability in dynamic environments.

[0098] Example 2

[0099] This embodiment provides an uplink sensing iterative optimization system based on a cellular-free architecture, applied to a distributed MIMO network, including:

[0100] Signal receiving unit: Deployed in a distributed access point (AP) array, each AP is equipped with a multi-antenna RF front end. It is configured to receive pilot signals and data signals transmitted by users; output initial channel estimation results and a received data matrix.

[0101] Processing Unit: Deployed in the central processing unit (CPU), it contains three interactive modules:

[0102] Data demodulation module: The improved D-AMP algorithm is implemented using FPGA, which performs iterative noise and signal updates and outputs the recovered communication symbols and demodulation residuals.

[0103] Multi-signal cooperative sensing module: Employs a GPU parallel computing architecture to implement the 2D-MUSIC algorithm. Inputs include pilot signals and communication symbols from the data demodulation module. The processing flow includes: constructing a joint channel matrix; calculating the angle spectrum and extracting the target angle; constructing a range spectrum based on the target angle and outputting the target range.

[0104] Dynamic channel reconstruction module: The input is the target position parameters output by the sensing module; the reconstruction logic is: according to the geometric channel model, the target parameters are mapped to the channel phase and amplitude; the output is: the reconstructed channel matrix.

[0105] Closed-loop control unit: dynamically feeds back the residual output from the data demodulation module to the multi-signal collaborative sensing module, adaptively adjusts the 2D-MUSIC search step size; performs historical weighted fusion of target position parameters and reconstructed channel; terminates iteration when the communication bit error rate is below the threshold and the change in target position parameters tends to stabilize.

[0106] In summary, this invention addresses the performance bottleneck caused by the coupling of communication and sensing in non-cellular systems by proposing an uplink sensing iterative optimization method based on a multi-module feedback mechanism. By establishing a closed-loop iterative process between data demodulation, target sensing, and channel reconstruction, a collaborative enhancement mechanism is achieved where communication error drives sensing optimization and sensing results assist channel reconstruction. This solves the problems of high bit error rate and low sensing accuracy caused by independent processing of communication and sensing in existing technologies, improves the overall reliability and sensing resolution of the system, and has significant practical application value.

[0107] Any aspects of this invention not described in detail are well-known to those skilled in the art.

[0108] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. An uplink sensing iterative optimization method based on a cellular-free system, wherein in a distributed MIMO system, there are M access points (APs), each equipped with S antennas, and K single-antenna users. Users and targets are randomly distributed within the coverage area, and the communication channel includes one Loss-S path and T Non-LoS paths; each Non-LoS path consists of a two-segment path of user-target-access point; characterized in that... The method specifically includes the following steps: Step S1: Establish an uplink non-cellular integrated communication and sensing system model. The user sends pilot signals for initial channel estimation and sends data to construct the received signal model, thereby obtaining the basic observation data required for communication and sensing. Step S2: Using the initial channel estimation and received data signal, execute the three module operations sequentially in each iteration: Data demodulation module: Recovers data symbols using an improved D-AMP algorithm; Multi-signal collaborative sensing module: Using the recovered data and pilot estimation as input, it uses the 2D-MUSIC algorithm to estimate the target's angle and distance; Dynamic channel reconstruction module: Uses sensing parameters to reconstruct the current communication channel using a geometric channel model; Step S3: The data demodulation residual feedback is used for parameter search and adjustment in the multi-signal collaborative sensing module, so that the communication error is reversed to optimize the sensing accuracy. The three modules drive and correct each other in each round, and continue to iterate until the communication bit error rate and the changes in sensing parameters meet the preset convergence conditions, and finally output the optimization result.

2. The method for estimating the covariance matrix of a distributed MIMO system based on fingerprint positioning according to claim 1, characterized in that, Step S1 specifically includes: Step S101: Establish an uplink cellular-free integrated communication and sensing system model. Under the nth subcarrier, the channel between the mth AP and the kth user is... Where, λ m,t,k q represents the large-scale fading factor. t,m q k,t These are the direction vectors for the target, AP, and user, respectively, d m,k,t The distance information is given by Δf, the subcarrier spacing is given by c, the speed of light is given by T, and the number of scatterers in the channel is given by T. Step S102: User k sends orthogonal pilot signals. The m-th AP receives the pilot signal as follows: Among them, P p It is the pilot transmission power, s k The pilot symbol sent by the k-th user is z, where z is additive Gaussian noise; The initial channel estimation is performed using the LS method based on the pilot signal. Among them, P p It is the pilot transmission power, N m It is additive Gaussian noise; the initial channel estimation on N subcarriers is defined as... τ p It is the pilot length. It is the pilot sequence of the kth user; Step S103: User k sends data x k Under the nth subcarrier, the data signal received by the mth AP is Among them, P d Here, z is the data transmission power, z is the additive Gaussian noise, and the received data matrix on N subcarriers is defined as follows: in, It is a matrix containing distance information. It is an angle information matrix, Z∈C MS×N It is a noise matrix.

3. The uplink sensing iterative optimization method based on a non-cellular system according to claim 2, characterized in that, The improved D-AMP algorithm for the data demodulation module in step S2 includes the following iterative steps: (a) Noise update: Where η1 is the weighting factor for incorporating historical iteration noise, X j This is the noise result from the previous iteration. It is a matrix containing distance information. It is an angle information matrix; (b) Signal update: Where j is the number of runs of the improved D-AMP algorithm, and η2 is the weighting factor for fusing historical iteration signals. It is the noise update output of the k-th user in the j-th iteration.

4. The uplink sensing iterative optimization method based on a non-cellular system according to claim 2, characterized in that, In step S2, the 2D-MUSIC algorithm of the target perception module is executed as follows: (a) Constructing the joint channel matrix: in For pilot signal receiving matrix, This represents the matrix obtained by removing data symbols from the received information symbols through element-wise division, where Q represents the total number of symbols. (b) Angle spectrum estimation: 1) Calculation The covariance matrix is ​​obtained, and eigenvalue decomposition is performed to extract the noise subspace E0; 2) Construct the angular spectrum function: Where a(θ) is the angular direction vector; F -1 The angle corresponding to the t-th maximum value of (θ) is the estimated target angle value for the t-th time. 3) Design a normalized filter vector w for each target angle estimate. t Construct the channel matrix from this angle. pass Constructing the distance spectrum function: Where a(d) is the distance direction vector, U tN yes The noise subspace of the covariance matrix; 4) Obtain the target distance estimate 5. The uplink synergistic iterative optimization method based on a non-cellular system according to claim 2, characterized in that, Step S3 specifically includes: Step S301: Input the received data signal, initial channel estimation, and number of iterations into the iterative optimization method, and initialize the data symbols and target estimation parameters; Step S302: In the (l+1)th iteration, execute the three module operations sequentially: Sub-step 1: In the multi-signal collaborative sensing module, recover the data symbols obtained in the l-th iteration. The channel matrix for this iteration is constructed based on the initial channel estimation, and the 2D-music algorithm is used to estimate the target's orientation. and distance parameters Historical perception results We obtain a weighted fusion with the current estimation results. Sub-step 2: Utilize the target parameter estimation results in the dynamic channel reconstruction module. The location relationship between the user and the AP is used to reconstruct the path components in the geometric channel model, and the reconstructed channel is calculated. Furthermore, the reconstruction results are weighted historically to obtain... η4 is the weighting factor for reconstructing the channel by fusing historical iterations; Sub-step 3: In the data demodulation module, based on the currently obtained channel estimation... An improved D-AMP algorithm is adopted to obtain the demodulated symbol X through noise update and signal update. l+1 and demodulation residuals Furthermore, the demodulation residual is fed back to the multi-signal collaborative sensing module in the next iteration to dynamically adjust the search step size in 2D-music; Step S303: Terminate the iteration if the following conditions are met: ‖BER l+1 -BER l ‖≤∈ comm ,‖A l+1 -A l ‖≤∈ sens Where ∈ comm and ∈ sens These are the performance thresholds for communication and sensing, respectively. If they are not met, repeat step S302.

6. An uplink sensing iterative optimization system based on a cellular-free architecture, applied to a distributed MIMO network, characterized in that... include: Signal receiving unit: configured to receive pilot signals and data signals sent by the user; Output the initial channel estimation results and the received data matrix; Processing unit: Contains three interactive modules: Data demodulation module: It uses an improved compressed sensing algorithm to recover communication symbols and outputs demodulation residuals; Multi-signal collaborative sensing module: fuses pilot signals and demodulated symbols, and extracts target position parameters through a two-dimensional spectrum estimation algorithm; Dynamic channel reconstruction module: maps target position parameters to a geometric channel model to reconstruct the communication channel; Closed-loop control unit: dynamically feeds back the residual output from the data demodulation module to the multi-signal collaborative sensing module to adjust the target search accuracy; performs historical weighted fusion of target position parameters and reconstructed channel; terminates the iteration when the communication bit error rate is below the threshold and the change in target position parameters tends to stabilize.

Images