Multi-node concurrent transmission signal superposition diversity method

Abstract

The invention discloses a multi-node concurrent transmission signal superposition diversity method. The method comprises the following steps: S1, generating a Turbo coding sequence by adopting a Turbo encoder; s2, applying phase disturbance to each Turbo coding sequence; s3, performing non-orthogonal superposition in an air channel to form a complex baseband signal; s4, constructing an equivalent observation model; s5, modeling the sending symbol matrix into a sparse matrix to obtain a symbol estimation matrix; s6, inputting the symbol estimation matrix into a Turbo decoder, executing iterative decoding by adopting an improved Max-Log-MAP algorithm, and outputting a bit estimation sequence; s7, completing initial phase estimation by using a Zadoff-Chu lead code, and outputting an estimated phase sequence through a Kalman filter; and S8, dynamically updating the equivalent observation model based on the estimated phase sequence, and completing parallel computing and scheduling. According to the method, Turbo coding, sparse reconstruction and phase disturbance modeling are fused, and multi-node non-orthogonal concurrent transmission is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of communication and signal processing, and particularly to a multi-node concurrent transmission signal superposition diversity method. Background Art

[0002] With the rapid development of wireless communication systems and Internet of Things applications, multi-node concurrent access and non-orthogonal transmission technologies have become key research directions for improving system capacity and spectrum utilization. Traditional orthogonal multiple access methods such as TDMA, FDMA, and OFDMA have problems of low resource allocation efficiency and high scheduling complexity in high-density connection scenarios. For this reason, non-orthogonal multiple access and sparse multiple access and other schemes have gradually become candidate technologies to replace orthogonal access. In such non-orthogonal systems, multiple users or nodes can superimpose and transmit signals on the same time-frequency resource, and the receiving end needs to recover each user's signal through a joint detection algorithm. However, due to the undifferentiated superposition of transmitted signals in the air, the receiving end faces extremely high interference cancellation complexity and reliable decoding challenges.

[0003] Currently, multi-node concurrent transmission systems often adopt precoding, multi-user detection (MUD), or iterative detection and decoding and other methods to mitigate non-orthogonal interference. The precoding scheme depends on the global channel state information at the sending end and is difficult to adapt to large-scale node scenarios; the computational complexity of the MUD scheme grows exponentially with the number of nodes, and its performance fluctuates greatly under strong interference or sparse connection conditions. To improve decoding reliability and computational efficiency, some systems introduce sparse modeling methods, regard the multi-user transmitted signals as a sparse matrix, and implement symbol reconstruction through compressive sensing, matching pursuit, and approximate message passing algorithms. Although such methods have certain performance advantages in theory, there are problems such as limited recovery accuracy, cumulative reconstruction error, and serious performance degradation when channel estimation is inaccurate in actual deployment.

[0004] On the other hand, to improve the robustness and diversity gain of the system, some communication systems introduce phase perturbation or spread spectrum modulation technologies, and improve the signal separation ability by introducing a pseudo-random rotation sequence into the transmitted signal. However, most of the existing methods stay at the static perturbation level, fail to achieve a dynamically adjustable phase structure, and the receiving end's dependence on the perturbation estimation relies on blind algorithms or pilot-assisted estimation, making it difficult to track the phase drift under channel changes, resulting in a large mismatch between the receiving model and the true signal, thereby affecting symbol detection and subsequent decoding performance.

[0005] In serial iterative structures such as Turbo decoding, due to the lack of joint modeling of signal structure and perturbation information, over-enhanced confidence or transmission distortion is likely to occur during the extrinsic information update process, resulting in the bit error rate tending to plateau in the high signal-to-noise ratio range, which limits the upper limit of system performance. Although some studies have attempted to simplify the decoding process through Max-Log-MAP and introduce a scaling factor to adjust the confidence information, a complete end-to-end system design has not been formed yet. In particular, there is a lack of an effective coupling mechanism among multiple sub-modules such as sparse reconstruction, Turbo decoding, phase estimation, and model feedback.

[0006] Therefore, how to provide a multi-node concurrent transmission signal superposition diversity method is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0007] An object of the present invention is to propose a multi-node concurrent transmission signal superposition diversity method. The present invention integrates Turbo coding and an improved Max-Log-MAP decoding method, introduces phase perturbation driven by Kasami codes and Zadoff-Chu preamble-assisted estimation, combines sparse modeling and approximate message passing algorithm, realizes non-orthogonal concurrent superposition transmission of multi-node signals in the air, and dynamically constructs an updatable observation model at the receiving end, which has the advantages of strong interference suppression, accurate symbol recovery, high decoding reliability, excellent computational efficiency, and strong system scalability, and is applicable to large-scale dense access scenarios.

[0008] A multi-node concurrent transmission signal superposition diversity method according to an embodiment of the present invention includes the following steps: S1. Use a Turbo encoder to encode the original bit stream of each node, and adjust it by puncturing according to the target code rate to generate a Turbo coding sequence; S2. Apply phase perturbation to each Turbo coding sequence based on a pseudo-random sequence, where the phase perturbation is driven by a binary sequence composed of Kasami codes to generate a phase modulation sequence; S3. Modulate and synthesize the Turbo coding sequence and the phase modulation sequence of each node, and multiple nodes transmit synchronously, and perform non-orthogonal superposition in the air channel to form a complex baseband signal; S4. The receiving end constructs an equivalent observation model according to the complex baseband signal, and the equivalent observation model is expressed as that the received signal matrix is composed of the product of the channel response matrix, the phase perturbation matrix, and the transmitted symbol matrix plus a noise term; S5. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix, construct an over-complete dictionary, and use the approximate message passing algorithm for sparse reconstruction to obtain a symbol estimation matrix; S6. Input the symbol estimation matrix into the Turbo decoder, perform iterative decoding using the improved Max-Log-MAP algorithm, introduce a scaling factor to adjust the confidence information update process, and output a bit estimation sequence; S7. Use the Zadoff-Chu preamble to complete the initial phase estimation, input the bit estimation sequence as soft information, and output an estimated phase sequence through a Kalman filter; S8. Dynamically update the equivalent observation model based on the estimated phase sequence, and use a pipeline structure and an AXI-Stream interface to complete parallel computing and scheduling.

[0009] Optionally, the specific steps of S1 include: S11. Configure a set of Turbo encoders for each node, and each Turbo encoder includes a first coding branch, a second coding branch, and an interleaver; S12. Input the original bit stream into the first coding branch and the interleaver. The first coding branch directly receives the input of the original bit stream, and the interleaver permutes the original bit stream according to the interleaving rule. The interleaving rule adopts the block interleaving method, fills the input bit stream row by row according to the set row-column mapping relationship, and then outputs it column by column to form an interleaved bit stream; S13. Input the interleaved bit stream into the second coding branch. The first coding branch and the second coding branch respectively perform recursive convolutional coding on the corresponding received bit streams, and output a first coding sequence and a second coding sequence; S14. Merge the original bit stream with the first coding sequence and the second coding sequence to form a Turbo-coded output bit stream; S15. Perform puncturing on the Turbo-coded output bit stream according to the redundancy requirement corresponding to the preset target code rate, delete redundant bits, and generate a Turbo-coded sequence.

[0010] Optionally, the specific steps of S2 include: S21. Generate a set of Kasami code sequences of length for each node , where represents the node number, represents the symbol index, and the Kasami code sequence is a periodic pseudo-random binary sequence; S22. Set the chip period and the phase perturbation step , where the chip period is the time duration corresponding to each symbol, and the phase perturbation step is a constant value; S23. Map each symbol to the corresponding time interval , maintaining a constant phase perturbation value within the time interval ; S24. Construct a continuous-time phase perturbation function for the node : ; wherein, represents the instantaneous phase rotation amount applied to the node at the time point , is the unit rectangular function, defined as taking the value of 1 when and 0 otherwise, used to limit that each perturbation value is constantly valid within the chip period; S25. Based on the continuous-time phase perturbation function, construct a complex exponential phase modulation function for the node : ; wherein, represents the phase modulation factor of the node at the time point , represents the imaginary unit, satisfying and is used to represent the rotation operation in the complex plane; S26. Based on the complex exponential phase modulation function, output the phase modulation sequence of each node.

[0011] Optionally, the S3 specifically includes: S31. Obtain the Turbo coding sequence of each node and the corresponding phase modulation sequence, multiply each symbol of the Turbo coding sequence by the corresponding phase modulation factor in the phase modulation sequence to generate a phase modulation signal sequence; S32. Perform digital-to-analog conversion on the phase modulation signal sequence to convert it into an analog baseband signal for wireless transmission; S33. Through the multi-node time synchronization mechanism, perform synchronization control on all nodes so that all nodes simultaneously transmit their respective analog baseband signals within the unified transmission time slot; S34. In the air channel, the analog baseband signals transmitted by all nodes are superimposed in a non-orthogonal form on the receiving side to form a complex baseband signal.

[0012] Optionally, the S4 specifically includes: S41. The receiving end receives the complex baseband signals synchronously transmitted by multiple nodes in real time and discretizes them in the time and space dimensions to form a received signal matrix , where each column corresponds to a symbol time slot and each row corresponds to the observed data of a receiving antenna channel; S42. Construct an equivalent observation model and represent the received signal matrix as: ; where represents the channel response matrix with a dimension of , indicating the complex channel gain from transmitting nodes to receiving antennas; represents the phase perturbation matrix, which is a complex diagonal matrix with a dimension of , and the diagonal elements are the phase modulation factors at the corresponding time points of each node; represents the transmitted symbol matrix with a dimension of , indicating the total number of complex modulation symbols transmitted by each node within a time slot; represents the noise matrix, which contains the complex Gaussian white noise components received by each channel.

[0013] Optionally, the S5 specifically includes: S51. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix. The sparse matrix means that within any time slot, no more than 10% of the total number of all nodes are in the active state, that is, the corresponding symbol elements are non-zero, and the remaining symbol elements are zero; S52. Perform a multiplication operation on the channel response matrix and the phase perturbation matrix to construct an over-complete dictionary matrix required for sparse reconstruction. The over-complete dictionary matrix represents the complex linear mapping relationship from the transmitted symbol space to the received signal space; S53. Establish the objective function of the sparse reconstruction problem: such that: ; where represents the L1 norm of the transmitted symbol matrix, represents the minimization operation, represents the received signal matrix, represents the over-complete dictionary matrix, represents the square of the L2 norm, represents the preset tolerance threshold; S54. Use the approximate message passing algorithm to iteratively solve the sparse reconstruction problem, specifically including: initializing the symbol estimation matrix and the observation residual, adjusting the symbol estimation based on the feedback of the previous round of observation residual in each round of iteration, updating the posterior mean of each symbol element by combining the noise variance and the Laplace sparse prior, and terminating the iteration when the estimation difference between two consecutive iterations is less than the set difference threshold or the maximum number of iterations is reached; S55. Output the symbol estimation matrix solved by the approximate message passing algorithm.

[0014] Optionally, S6 specifically includes: S61. Input the transmitted symbol estimation matrix as soft information into the Turbo decoder corresponding to each node. The Turbo decoder includes a first decoding branch, a second decoding branch, and an interleaver, forming an iterative structure; S62. In each iteration, use the improved Max-Log-MAP algorithm to calculate the log-likelihood ratio of the information bits , where the improved Max-Log-MAP algorithm constructs a metric difference according to the forward recursion variable , the backward recursion variable , the branch metric function , and the state transition path of the received information: ; Among them, and respectively represent the start state and the end state of the state transition; S63. The branch metric function represents the log-likelihood metric of the path transferred from the start state to the end state at the symbol time : ; Among them, represents the received symbol at the th moment, represents the conditional probability of receiving when the input information bit is and the path is , represents the prior probability of the input information bit ; S64. When calculating the extrinsic information, introduce a scaling factor , and perform scaling correction and update on the extrinsic information generated in each iteration; S65. Rearrange the bit positions of the extrinsic information generated by the first decoding branch through the interleaver and input it into the second decoding branch. The second decoding branch uses the extrinsic information as the prior probability, combines the observed signal received by itself to perform path recursion operations and metric updates, obtains new extrinsic information, and transmits the new extrinsic information generated by the second decoding branch back to the first decoding branch through the de-interleaver as the prior input for the next round of iteration. The two decoding branches alternate until the maximum number of iterations is reached or the change in the bit log-likelihood ratio between two consecutive rounds is less than the preset change threshold; S66. Output the bit estimation sequence after the final iteration of the Turbo decoder as the restored original information bit stream.

[0015] Optionally, S7 specifically includes: S71. Embed a preamble composed of Zadoff-Chu sequences in each node's transmitted signal frame to assist the receiving end in initial phase estimation. S72. The receiving end extracts the Zadoff-Chu preamble from the received complex baseband signal and completes the initial phase offset estimation for each node based on the locally known sequence. S73. Use the bit estimation sequence as soft information input, and combine the phase evolution behavior of the received signal among multiple symbols to construct a dynamic observation data sequence. S74. Based on the initial phase offset estimation of each node and the dynamic observation data sequence, use a Kalman filter for recursive estimation of phase perturbation, real-time track phase drift and carrier frequency offset, and gradually correct the estimated value. S75. Output the complete estimated phase sequence of each node.

[0016] Optionally, S8 specifically includes: S81. The receiving end obtains the estimated phase sequence corresponding to each node, and the estimated phase sequence represents the phase rotation change experienced by the node at each symbol time point. S82. Update the phase perturbation matrix in the equivalent observation model according to the estimated phase sequence, and replace the diagonal elements corresponding to each node in the phase perturbation matrix with complex modulation factors constructed according to the estimated phase sequence. S83. Construct a new equivalent observation model based on the updated phase perturbation matrix to describe the mapping relationship between the current received signal and the transmitted symbol. S84. Adopt a pipeline structure to improve the parallelism of data processing, and construct a data transmission path through the AXI-Stream interface to complete scheduling, and automatically update the equivalent observation model.

[0017] The beneficial effects of the present invention are: (1) By introducing Kasami code-driven pseudo-random phase perturbation into the transmitted signals of each node, differential modulation of physical layer non-orthogonal superposition is realized, the separability of signals during air superposition is improved, and the interference coupling degree between multiple nodes is reduced. (2) Adopt the joint design of Turbo coding and Max-Log-MAP decoding, and combine the dynamic adjustment mechanism of soft information and scaling factor to effectively improve the decoding convergence and the bit error performance under low signal-to-noise ratio. (3) Model the received symbols as a sparse matrix, and complete low-complexity sparse reconstruction through the approximate message passing algorithm to improve the symbol estimation accuracy in the case of partial node activation and complex channel conditions; (4) Introduce Zadoff-Chu preamble and Kalman filter to achieve recursive estimation of phase perturbation and frequency offset compensation, enhance the adaptive ability of the receiving end to dynamic phase changes, and improve the system robustness; (5) Construct a dynamically updatable equivalent observation model, and combine the pipeline structure and AXI-Stream interface to achieve module-level parallel computing, optimize the system throughput rate and deployability, and form a multiple diversity gain mechanism in the space, time, and modulation domains, which is applicable to large-scale distributed wireless communication scenarios. Description of the Drawings

[0018] The drawings are used to provide further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. In the drawings: Figure 1 is the overall flowchart of a multi-node concurrent transmission signal superposition diversity method proposed by the present invention; Figure 2 is the structural schematic diagram of the Turbo encoder of a multi-node concurrent transmission signal superposition diversity method proposed by the present invention; Figure 3 is the Max-Log-MAP decoding flowchart based on the Turbo decoder structure of a multi-node concurrent transmission signal superposition diversity method proposed by the present invention. Detailed Embodiments

[0019] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0020] Refer to Figures 1-3 , a multi-node concurrent transmission signal superposition diversity method, including the following steps: S1. Encode the original bit stream of each node using a Turbo encoder, and adjust it by puncturing according to the target code rate to generate a Turbo coding sequence; S2. Apply phase perturbation to each Turbo coding sequence based on a pseudo-random sequence, and the phase perturbation is driven by a binary sequence composed of Kasami codes to generate a phase modulation sequence; S3. Modulate and synthesize the Turbo coding sequence of each node with the phase modulation sequence, and multiple nodes transmit synchronously, and perform non-orthogonal superposition in the air channel to form a complex baseband signal; S4. The receiving end constructs an equivalent observation model based on the complex baseband signal. The equivalent observation model is expressed as that the received signal matrix is composed of the product of the channel response matrix, the phase perturbation matrix and the transmitted symbol matrix plus a noise term; S5. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix, construct an over-complete dictionary, and use the approximate message passing algorithm for sparse reconstruction to obtain a symbol estimation matrix; S6. Input the symbol estimation matrix into the Turbo decoder, use the improved Max-Log-MAP algorithm to perform iterative decoding, introduce a scaling factor to adjust the confidence information update process, and output a bit estimation sequence; S7. Use the Zadoff-Chu preamble to complete the initial phase estimation, input the bit estimation sequence as soft information, and output an estimated phase sequence through a Kalman filter; S8. Dynamically update the equivalent observation model based on the estimated phase sequence, and use a pipeline structure and an AXI-Stream interface to complete parallel computing and scheduling.

[0021] This method realizes the reliable separation and restoration of multi-node data in an air non-orthogonal superposition environment by constructing an end-to-end multi-node concurrent transmission and receiving decoding system, integrating Turbo coding, phase perturbation, sparse modeling and dynamic observation update technologies. Through the soft information-driven Turbo iterative structure and the perturbation estimation of the Kalman filter, the adaptability of the system to phase changes and frequency offset dynamic channel conditions is enhanced. The overall architecture has high parallelism and deployability, is suitable for scenarios such as the Internet of Things and large-scale access, and has good robustness and high channel capacity utilization.

[0022] In this embodiment, the specific content of S1 includes: S11. Configure a set of Turbo encoders for each node. The Turbo encoder includes a first coding branch, a second coding branch and an interleaver; S12. Input the original bit stream into the first coding branch and the interleaver. The first coding branch receives the direct input of the original bit stream, and the interleaver permutes the original bit stream according to the interleaving rule. The interleaving rule adopts a block interleaving method, and fills the input bit stream by rows and outputs by columns according to the set row-column mapping relationship to form an interleaved bit stream; S13. Input the interleaved bit stream into the second coding branch. The first coding branch and the second coding branch respectively perform recursive convolutional coding on the corresponding received bit streams, and output a first coding sequence and a second coding sequence; S14. Merge the original bit stream with the first coding sequence and the second coding sequence to form a Turbo coding output bit stream; S15. Punch process the output bitstream of the Turbo coding according to the redundancy requirement corresponding to the preset target code rate, delete the redundant bits, and generate a Turbo coding sequence.

[0023] By adopting a Turbo coding structure composed of two recursive convolutional encoders and an interleaver, the coding redundancy and structural symmetry of the original information are effectively improved. The introduction of the block interleaver can break the correlation between bits and enhance the anti-interference ability against burst errors. This structure has good soft information traceability and decomposability in iterative decoding, provides stable redundancy support for the recovery process at the receiving end, effectively reduces the bit error rate, and enhances the channel coding gain.

[0024] In this embodiment, the specific steps of S2 include: S21. Generate a set of Kasami code sequences with a length of for each node , where represents the node number, represents the symbol index, and the Kasami code sequence is a periodic pseudo-random binary sequence; S22. Set the chip period and the phase perturbation step size , where the chip period is the time duration corresponding to each symbol, and the phase perturbation step size is a constant value; S23. Map each symbol to the corresponding time interval , and maintain a constant phase perturbation value within the time interval; S24. Construct the continuous-time phase perturbation function of node : ; where, represents the instantaneous phase rotation amount applied to node at time point , is the unit rectangular function, defined as taking the value of 1 when , and taking the value of 0 otherwise, which is used to limit the constancy and effectiveness of each perturbation value within the chip period; S25. Based on the continuous-time phase perturbation function, construct the complex exponential phase modulation function of node : where, represents the phase modulation factor of node at time point , denotes the imaginary unit and satisfies , and is used to represent the rotation operation in the complex plane; S26. Based on the complex exponential phase modulation function, output the phase modulation sequence of each node.

[0025] The phase perturbation mechanism adopts a binary pseudo-random sequence driven by Kasami codes, combines a continuous-time perturbation function with a complex exponential modulation expression, enables each node to have a unique identifier under the condition of co-frequency transmission, thereby improving the signal discrimination. A modulation sequence with time consistency is formed through constant-step phase perturbation, creating a prior basis for interference suppression and decoding separation at the receiving end, and enhancing the anti-interference ability and non-orthogonal decoupling ability of the multi-node system.

[0026] In this embodiment, the specific steps of S3 are as follows: S31. Obtain the Turbo coding sequence of each node and the corresponding phase modulation sequence, multiply each symbol of the Turbo coding sequence by the corresponding phase modulation factor in the phase modulation sequence to generate a phase modulation signal sequence; S32. Perform digital-to-analog conversion on the phase modulation signal sequence to convert it into an analog baseband signal for wireless transmission; S33. Through the multi-node time synchronization mechanism, synchronize and control all nodes so that all nodes simultaneously transmit their respective analog baseband signals within the unified transmission time slot; S34. In the air channel, the analog baseband signals transmitted by all nodes are superimposed in a non-orthogonal form on the receiving side to form a complex baseband signal.

[0027] At the transmitting side, through the complex multiplication of Turbo coding symbols and phase modulation factors, physical layer modulation mapping of the signal is realized, and an analog baseband signal is formed after digital-to-analog conversion. The multi-node time synchronization mechanism ensures the consistency of the signal at the transmitting side, and a stable non-orthogonal superposition structure is formed in the air. This method realizes multi-user concurrency without the need for frequency division and code division, improves the system spectrum efficiency, and reduces the scheduling complexity.

[0028] In this embodiment, the specific steps of S4 are as follows: S41. The receiving end receives the complex baseband signals synchronously transmitted by multiple nodes in real time, and discretizes them in the time and space dimensions to form a received signal matrix , where each column corresponds to a symbol time slot, and each row corresponds to the observed data of a receiving antenna channel; S42. Construct an equivalent observation model, and represent the received signal matrix as: ; Among them, represents the channel response matrix, and the dimension is , representing the complex channel gain from transmitting nodes to receiving antennas; represents the phase perturbation matrix, which is a complex diagonal matrix with a dimension of , and the diagonal elements are the phase modulation factors of each node at the corresponding time points; represents the transmitted symbol matrix, with a dimension of , represents the total number of complex modulation symbols transmitted by each node within a time slot; represents the noise matrix, which contains the complex Gaussian white noise components received by each channel.

[0029] The equivalent observation model is expressed as that the received signal matrix is composed of the product of the channel response matrix, the phase perturbation matrix and the transmitted symbol matrix plus the noise term, realizing the unified abstract modeling of the actual channel and the modulation mechanism. The equivalent observation model has the characteristics of being discretizable, extensible and computable, providing high-fidelity input conditions for sparse reconstruction and decoding. At the same time, the matrix form can be compatible with the multi-antenna reception structure and has natural space diversity ability.

[0030] In this embodiment, the S5 specifically includes: S51. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix. The sparse matrix means that in any time slot, no more than 10% of the total number of all nodes are in the active state, that is, the corresponding symbol elements are non-zero, and the rest of the symbol elements are zero; S52. Perform a product operation on the channel response matrix and the phase perturbation matrix to construct an over-complete dictionary matrix required for sparse reconstruction. The over-complete dictionary matrix represents the complex linear mapping relationship from the transmitted symbol space to the received signal space; S53. Establish the objective function of the sparse reconstruction problem: such that: ; where represents the L1 norm of the transmitted symbol matrix, represents the minimization operation, represents the received signal matrix, represents the over-complete dictionary matrix, represents the square of the L2 norm, represents the preset tolerance threshold; S54. Use the approximate message passing algorithm to iteratively solve the sparse reconstruction problem, specifically including: initializing the symbol estimation matrix and the observation residual, in each iteration, adjusting the symbol estimation based on the feedback of the previous iteration's observation residual, updating the posterior mean of each symbol element by combining the noise variance and the Laplace sparse prior, and terminating the iteration when the estimation difference between two consecutive iterations is less than the set difference threshold or the maximum number of iterations is reached; S55. Output the symbol estimation matrix obtained by solving with the approximate message passing algorithm.

[0031] By sparsely modeling the transmitted symbol matrix and introducing an overcomplete dictionary and a sparse reconstruction solution framework, the computational dimension during decoding is significantly reduced. The approximate message passing algorithm has characteristics such as low complexity, fast convergence, and good robustness, and is suitable for dense access scenarios with extremely low node activity rates. It has high reconstruction accuracy and a clear iteration termination mechanism, avoiding the risks of model oscillation or misestimation, thus ensuring the effective input for subsequent Turbo decoding.

[0032] In this embodiment, the specific content of S6 includes: S61. Input the transmitted symbol estimation matrix as soft information into the Turbo decoder corresponding to each node. The Turbo decoder includes a first decoding branch, a second decoding branch, and an interleaver, forming an iterative structure; S62. In each iteration, use the improved Max-Log-MAP algorithm to calculate the log-likelihood ratio of the information bit . The improved Max-Log-MAP algorithm constructs a metric difference according to the forward recursion variable , the backward recursion variable , the branch metric function and the state transition path of the received information: ; where and respectively represent the starting state and the ending state of the state transition; S63. The branch metric function represents the log-likelihood metric of the path from the starting state to the ending state at the symbol time : ; where represents the received symbol at the th moment, represents that when the input information bit is and the path is , the received The conditional probability represents the prior probability of the input information bit ; S64. While calculating the extrinsic information, introduce a scaling factor , and perform scaling correction and update on the extrinsic information generated in each iteration; S65. Rearrange the bit positions of the extrinsic information generated by the first decoding branch through the interleaver and input it into the second decoding branch. The second decoding branch uses the extrinsic information as the prior probability, combines the observed signal received by itself, performs path recursion operation and metric update, obtains new extrinsic information, and transmits the new extrinsic information generated by the second decoding branch back to the first decoding branch through the deinterleaver as the prior input for the next round of iteration. The two decoding branches alternate until the maximum number of iterations is reached or the change in the bit log-likelihood ratio between two consecutive rounds is less than the preset change threshold; S66. Output the bit estimation sequence after the final iteration of the Turbo decoder as the restored original information bit stream.

[0033] The improved Max-Log-MAP algorithm implements a dual-branch alternating iteration structure in the Turbo decoder. The soft information is transmitted between branches through the interleaver and deinterleaver, continuously optimizing the prior probability estimation. Introducing a scaling factor to control the update intensity can prevent the confidence information from being oversaturated and improve the decoding stability. The overall structure can converge to the optimal solution within a limited number of iterations, significantly improving the information recovery ability of the system under weak channel conditions.

[0034] In this embodiment, the specific steps of S7 include: S71. Embed a preamble composed of Zadoff-Chu sequences in each node's transmitted signal frame to assist the receiving end in initial phase estimation; S72. The receiving end extracts the Zadoff-Chu preamble from the received complex baseband signal and completes the initial phase offset estimation of each node based on the locally known sequence; S73. Input the bit estimation sequence as soft information, and construct a dynamic observation data sequence by combining the phase evolution behavior of the received signal among multiple symbols; S74. Based on the initial phase offset estimation of each node and the dynamic observation data sequence, use the Kalman filter to perform recursive estimation of the phase perturbation, real-time track the phase drift and carrier frequency offset, and gradually correct the estimated value; S75. Output the complete estimated phase sequence of each node.

[0035] The precise initial phase estimation is achieved by utilizing the good autocorrelation property of the Zadoff-Chu preamble. Combining the bit soft information with the received signal state and performing recursive update through Kalman filtering can accurately track the phase perturbation and frequency offset variation in the node's transmitted signal. This estimation structure has good response ability in a dynamic environment, provides a reliable basis for the update of the equivalent observation model, and improves the robustness of the entire decoding chain.

[0036] In this embodiment, S8 specifically includes: S81. The receiving end obtains the estimated phase sequence corresponding to each node, and the estimated phase sequence represents the phase rotation change experienced by the node at each symbol time point; S82. Update the phase perturbation matrix in the equivalent observation model according to the estimated phase sequence, and replace the diagonal elements corresponding to each node in the phase perturbation matrix with the complex modulation factors constructed according to the estimated phase sequence; S83. Construct a new equivalent observation model based on the updated phase perturbation matrix to describe the mapping relationship between the current received signal and the transmitted symbol; S84. Adopt a pipeline structure to improve the parallelism of data processing, and construct a data transmission path through the AXI-Stream interface to complete scheduling and automatically update the equivalent observation model.

[0037] Based on the estimated phase sequence, the phase perturbation matrix in the equivalent observation model is updated in real time, and an equivalent receiving channel mapping reflecting the current true state is constructed. The pipeline structure and the AXI-Stream interface support high-concurrency operations, improving the system throughput and execution efficiency. The model update mechanism has modularity and scalability, provides efficient decoupling and decoding capabilities for a large-scale node environment, and has good potential for engineering implementation.

[0038] Example 1: To verify the feasibility of the present invention in implementation, the present invention is applied to a certain simulated multi-node communication platform to construct a representative wireless transmission scenario. This platform includes multiple heterogeneous transmitting nodes and a multi-antenna receiving end. The system simulates the temporary deployment situation without a fixed base station to address the actual communication problems of high interference and frequent dynamic changes in the channel in the city.

[0039] In test scenarios such as disaster emergency, unmanned combat networks, and / or edge computing environments, multiple terminal devices need to upload critical data to the central node simultaneously. In this scenario, traditional orthogonal multiple access methods are difficult to adapt to scenarios with high concurrency and resource constraints. They have low spectral efficiency and high requirements for synchronization accuracy, resulting in an increase in the bit error rate and frequent system bottlenecks. Therefore, the present invention designs a multi-node synchronous transmission mechanism, introduces Turbo coding to enhance the redundancy ability, and at the same time achieves signal decorrelation of nodes through phase perturbation driven by Kasami codes. Combining sparse modeling and message passing algorithms, joint recovery of multi-node signals is achieved at the receiving end.

[0040] In the simulation test, each node encodes the bit stream through an independent Turbo encoder, and transmits it after superimposing the phase perturbation driven by Kasami codes. The receiving end constructs an equivalent observation model, uses a sparse reconstruction algorithm to estimate the symbol matrix, and then performs Turbo decoding using an improved Max-Log-MAP algorithm. Precise phase tracking and frequency offset compensation are completed through Kalman filtering. The entire processing flow is deployed on a programmable platform, and parallel processing is achieved using a pipeline structure and an AXI-Stream bus architecture to ensure high-throughput data throughput.

[0041] It is observed from five consecutive rounds of tests that when the number of nodes is fixed at 8 and the channel SNR covers the range of 6.5 dB to 14.5 dB, the system exhibits good robustness. The number of iterations of Turbo decoding is controlled within the range of 4 to 7 times, and the average bit error rate is as low as 0.867% at the lowest and 1.550% at the highest, significantly superior to traditional non-orthogonal communication schemes. The symbol recovery accuracy rate remains between 97.13% and 99.20%, and the average frequency offset compensation error is less than 0.05 rad, indicating that Kalman filtering can effectively suppress the phase distortion caused by multipath effects and frequency offsets. At the same time, the total system decoding delay is controlled within 1622 milliseconds, meeting the requirements of low-latency communication.

[0042] Table 1 Summary of test results of the multi-node Turbo decoding scheme

[0043] The test results in Table 1 above fully demonstrate that the method proposed by the present invention can achieve highly reliable non-orthogonal signal recovery in a multi-node asynchronous environment, and has excellent bit error suppression ability and decoding efficiency, effectively complementing and upgrading the traditional multiple access communication mode, and having strong application and promotion value.

[0044] In this embodiment, a multi-node concurrent signal superposition diversity scheme based on the combination of Turbo coding and sparse reconstruction is constructed, which significantly improves the transmission reliability and decoding performance in a non-orthogonal superposition environment. Experimental results show that under the test conditions of 8 nodes and medium-low average channel SNR, the proposed scheme of the present invention achieves a symbol recovery accuracy of over 97% while maintaining a low bit error rate (BER as low as 0.86%), effectively compensating for the performance degradation caused by frequency offset and phase perturbation. In addition, by introducing the soft information alternation mechanism and phase estimation feedback in the Max-Log-MAP decoding structure, the present invention can obtain a converged and stable bit estimation result within a limited number of iterations, meeting the requirements of low latency and high robustness in practical communication systems. This embodiment fully demonstrates the diversity effect in multi-node cooperative transmission, combines the strong error correction ability of Turbo coding and the decoupling advantage brought by sparse modeling, and constructs a highly scalable and reasonably computationally resource-consuming receiving and processing process.

[0045] As described above, the above is only the preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A multi-node concurrent transmission signal superposition diversity method, characterized in that It includes the following steps: S1. Use a Turbo encoder to encode the original bit stream of each node, and adjust it by puncturing according to the target code rate to generate a Turbo coding sequence; S2. Apply phase perturbation to each Turbo coding sequence based on a pseudo-random sequence, where the phase perturbation is driven by a binary sequence composed of Kasami codes to generate a phase modulation sequence; S3. Modulate and synthesize the Turbo coding sequence of each node with the phase modulation sequence, and multiple nodes transmit synchronously, and perform non-orthogonal superposition in the air channel to form a complex baseband signal; S4. The receiving end constructs an equivalent observation model according to the complex baseband signal, and the equivalent observation model is expressed as that the received signal matrix is composed of the product of the channel response matrix, the phase perturbation matrix and the transmitted symbol matrix plus a noise term; S5. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix, construct an over-complete dictionary, and use the approximate message passing algorithm for sparse reconstruction to obtain a symbol estimation matrix; S6. Input the symbol estimation matrix into a Turbo decoder, and use an improved Max-Log-MAP algorithm to perform iterative decoding, introduce a scaling factor to adjust the confidence information update process, and output a bit estimation sequence; S7. Use a Zadoff-Chu preamble to complete the initial phase estimation, input the bit estimation sequence as soft information, and output an estimated phase sequence through a Kalman filter; S8. Dynamically update the equivalent observation model based on the estimated phase sequence, and use a pipeline structure and an AXI-Stream interface to complete parallel computing and scheduling.

2. The multi-node concurrent transmission signal superposition diversity method according to claim 1, wherein, The specific content of S1 includes: S11. Configure a set of Turbo encoders for each node, and the Turbo encoder includes a first coding branch, a second coding branch and an interleaver; S12. Input the original bit stream into the first coding branch and the interleaver. The first coding branch receives the direct input of the original bit stream, and the interleaver permutes the original bit stream according to the interleaving rule. The interleaving rule adopts a block interleaving method, and fills the input bit stream row by row and outputs it column by column according to the set row-column mapping relationship to form an interleaved bit stream; S13. Input the interleaved bit stream into the second coding branch, and the first coding branch and the second coding branch respectively perform recursive convolutional coding on the corresponding received bit stream to output a first coding sequence and a second coding sequence; S14. Merge the original bit stream with the first coding sequence and the second coding sequence to form a Turbo coding output bit stream; S15. Perform puncturing processing on the Turbo coding output bit stream according to the redundancy requirement corresponding to the preset target code rate, delete redundant bits, and generate a Turbo coding sequence.

3. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that The specific content of S2 includes: S21. Generate a set of Kasami code sequences of length for each node , where represents the node number, represents the symbol index, and the Kasami code sequence is a periodic pseudo-random binary sequence; S22. Set the chip period and the phase perturbation step size , where the chip period is the time duration corresponding to each symbol, and the phase perturbation step size is a constant value; S23. Map each symbol to the corresponding time interval and maintain a constant phase perturbation value within the time interval ; S24. Construction node Continuous-time phase perturbation function: ； Among them, represents the instantaneous phase rotation amount applied to the upper node at the time point is the unit rectangular function, defined as taking the value of 1 when and taking the value of 0 otherwise, and is used to limit that each perturbation value is constantly valid within the chip period; S25. Construct a complex exponential phase modulation function of the node based on the continuous-time phase perturbation function : ； Among them, represents the phase modulation factor of the upper node at the time point represents the imaginary unit, satisfying , and is used to represent the rotation operation in the complex plane; S26. Based on the complex exponential phase modulation function, output the phase modulation sequence of each node.

4. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that, The specific content of S3 includes: S31. Obtain the Turbo coding sequence of each node and the corresponding phase modulation sequence, multiply each symbol of the Turbo coding sequence by the corresponding phase modulation factor in the phase modulation sequence to generate a phase modulation signal sequence; S32. Perform digital-to-analog conversion on the phase modulation signal sequence to convert it into an analog baseband signal for wireless transmission; S33. Through the multi-node time synchronization mechanism, synchronously control all nodes so that all nodes simultaneously transmit their respective analog baseband signals within the unified transmission time slot; S34. In the air channel, the analog baseband signals transmitted by all nodes are superimposed in a non-orthogonal form on the receiving side to form a complex baseband signal.

5. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that, The specific content of S4 includes: S41. The receiving end receives in real time the complex baseband signals synchronously transmitted by multiple nodes, and discretizes them in the time and space dimensions to form a received signal matrix , where each column corresponds to a symbol time slot and each row corresponds to the observed data of a receiving antenna channel; S42. Construct an equivalent observation model and represent the received signal matrix as follows: ； Among them, represents the channel response matrix, with a dimension of , representing the complex channel gain from transmitting nodes to receiving antennas; represents the phase perturbation matrix, which is a complex diagonal matrix with a dimension of , and the diagonal elements are the phase modulation factors of each node at the corresponding time points; represents the transmitted symbol matrix, with a dimension of , representing the total number of complex modulation symbols transmitted by each node within a time slot; represents the noise matrix, which contains the complex Gaussian white noise components received by each channel.

6. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that, The specific content of S5 includes: S51. Model the transmitted symbol matrix in the equivalent observation model as a sparse matrix. The sparse matrix means that in any time slot, no more than 10% of the total number of all nodes are in the active state, that is, the corresponding symbol elements are non-zero, and the rest of the symbol elements are zero; S52. Perform a product operation on the channel response matrix and the phase perturbation matrix to construct an over-complete dictionary matrix required for sparse reconstruction. The over-complete dictionary matrix represents the complex linear mapping relationship from the transmitted symbol space to the received signal space; S53. Establish the objective function of the sparse reconstruction problem: such that: ; Among them, represents the L1 norm of the transmitted symbol matrix, represents the minimization operation, represents the received signal matrix, represents the overcomplete dictionary matrix, represents the square of the L2 norm, represents the preset tolerance threshold; S54. Use the approximate message passing algorithm to iteratively solve the sparse reconstruction problem, specifically including: initializing the symbol estimation matrix and the observation residual, adjusting the symbol estimation based on the feedback of the previous round of observation residual in each round of iteration, updating the posterior mean of each symbol element by combining the noise variance and the Laplacian sparse prior, and terminating the iteration when the estimation difference between two consecutive iterations is less than the set difference threshold or the maximum number of iterations is reached; S55. Output the symbol estimation matrix after being solved by the approximate message passing algorithm.

7. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that The specific content of S6 includes: S61. Input the transmitted symbol estimation matrix as soft information into the Turbo decoder corresponding to each node. The Turbo decoder includes a first decoding branch, a second decoding branch, and an interleaver, forming an iterative structure; S62. In each iteration, the improved Max-Log-MAP algorithm is used to calculate the log-likelihood ratio of the information bits wherein the improved Max-Log-MAP algorithm constructs a metric difference according to the forward recursion variable , the backward recursion variable , the branch metric function and the state transition path of the received information: ​ ； Among them, and respectively represent the start state and the end state of the state transition; S63. The branch metric function represents the log-likelihood metric of the path corresponding to the transition from the starting state to the ending state at the symbol time : ； Among them, represents the received symbol at moment, represents the conditional probability of receiving when the input information bit is and the path is ; represents the prior probability of the input information bit ; S64. Introduce a scaling factor while calculating the extrinsic information , and perform scaling correction and update on the extrinsic information generated in each iteration; S65. Rearrange the bit positions of the extrinsic information generated by the first decoding branch through the interleaver and then input it into the second decoding branch. The second decoding branch uses the extrinsic information as the prior probability, combines the observed signal received by itself to perform path recursion operation and metric update to obtain new extrinsic information, and transmits the new extrinsic information generated by the second decoding branch back to the first decoding branch through the de-interleaver as the prior input for the next round of iteration. The two decoding branches alternately execute until the maximum number of iterations is reached or the change in the bit log-likelihood ratio between two consecutive rounds is less than the preset change threshold; S66. Output the bit estimation sequence after the final iteration of the Turbo decoder as the recovered original information bit stream.

8. A multi-node concurrent transmission signal superposition diversity method according to claim 1, characterized in that The specific content of S7 includes: S71. Embed a preamble composed of Zadoff-Chu sequences in the signal frame transmitted by each node to assist the receiving end in performing initial phase estimation; S72. The receiving end extracts the Zadoff-Chu preamble from the received complex baseband signal and completes the initial phase offset estimation of each node based on the locally known sequence. S73. Input the bit estimation sequence as soft information, and construct a dynamic observation data sequence in combination with the phase evolution behavior between multiple symbols of the received signal. S74. Based on the initial phase offset estimation of each node and the dynamic observation data sequence, use the Kalman filter for recursive estimation of phase perturbation, track the phase drift and carrier frequency offset in real time, and gradually correct the estimated value. S75. Output the complete estimated phase sequence of each node.

9. A multi-node concurrent transmission signal superposition diversity method according to claim 1, wherein The specific content of S8 includes: S81. The receiving end obtains the estimated phase sequence corresponding to each node, and the estimated phase sequence represents the phase rotation change experienced by the node at each symbol time point. S82. Update the phase perturbation matrix in the equivalent observation model according to the estimated phase sequence, and replace the diagonal elements corresponding to each node in the phase perturbation matrix with the complex modulation factors constructed according to the estimated phase sequence. S83. Construct a new equivalent observation model based on the updated phase perturbation matrix to describe the mapping relationship between the current received signal and the transmitted symbol. S84. Adopt a pipeline structure to improve the parallelism of data processing, construct a data transmission path through the AXI-Stream interface to complete scheduling, and automatically update the equivalent observation model.

Images