A low channel state information overhead transceiver for a double selective fading channel and a method thereof

Abstract

The application provides a low channel state information (CSI) overhead transceiver for a double selective fading channel, including a transmitting end and a receiving end, wherein the transmitting end includes, when the CSI budget is limited, the receiving end can be provided with the minimum necessary identifiable information, so as to avoid detection drift and performance degradation caused by scale uncertainty, sign inversion or insufficient identifiability under weak CSI; and compared with a high density pilot structure, the auxiliary resource occupies a lower proportion, and is more in line with the low overhead design goal of engineering landing.

Description

Technical Field

[0001] This invention belongs to the field of communication technology and designs a low channel state information overhead transceiver for dual-select fading channels. Background Technology

[0002] In applications such as high-speed mobile, vehicle-to-everything (V2X), rail transit, drones, and low-Earth orbit satellites, wireless channels exhibit significant time-varying Doppler spread and frequency-selective multipath effects, forming typical dual-selective fading (DSF) channels. In such channel environments, traditional modulation and equalization based on orthogonal frequency division multiplexing (OFDM) typically rely on relatively accurate channel state information (CSI) to suppress inter-symbol interference (ISI) and inter-carrier interference (ICI). However, as mobile speed increases or Doppler spread intensifies, the frequency of CSI acquisition, tracking, and updating increases significantly, leading to a simultaneous increase in pilot overhead, guard interval overhead, and receiver computational complexity, thereby reducing effective throughput and system energy efficiency.

[0003] Orthogonal Time-Frequency Spatial Modulation (OTFS) transmits information in the time-frequency domain while carrying it in the delay-Doppler domain. This makes the DSF channel more conducive to structured modeling and processing in the delay-Doppler domain, thus it is considered suitable for reliable communication under high-speed mobile conditions. However, OTFS research generally points out that receiver performance largely depends on the quality of channel estimation and symbol detection: in practical receivers, insufficient CSI acquisition or large estimation errors will directly affect equalization and detection performance, becoming one of the core bottlenecks for the practical implementation of OTFS.

[0004] For the channel estimation problem of OTFS, existing technical approaches mainly include pilot-assisted delay-Doppler domain sparse estimation, structured matched pursuit, and model / data-driven estimation methods. Among them, delay-Doppler domain sparsity is a commonly used prior: under the assumption of a finite number of scatterers, the channel is characterized by a small number of path parameters (delay, Doppler, and complex gain), and sparse recovery or structured estimation can be used to reduce parameter dimensionality and estimation complexity.

[0005] However, pilot-assisted channel estimation is often accompanied by significant system overhead and engineering costs. On the one hand, pilot overhead tends to increase with maximum delay spread and maximum Doppler spread, and to suppress the interference of pilot leakage on data signals, zero-filling or guard intervals are often required around the pilot, further sacrificing effective resources. Related research and published literature commonly employ the engineering approach of "setting up a guard zone around the pilot to avoid leakage," which incurs additional rate loss. On the other hand, when the Doppler is fractional (off-grid Doppler), path energy spreads across the delay-Doppler grid, causing "strict sparsity" to degenerate into "quasi-sparseness," thus requiring a denser dictionary, more complex compensation, or higher pilot energy to maintain estimation accuracy. To address fractional delay / fractional Doppler, some works have proposed schemes such as superimposed pilots to reduce guard zone and rate loss, but a trade-off between pilot design and detection robustness is still necessary.

[0006] Besides pilot-assisted approaches, semi-blind / blind estimation and differential modulation-assisted channel processing have also attracted attention. Some studies utilize differential modulation or differential structures to reduce or eliminate the need for pilots and protected areas, thereby reducing terminal power consumption and improving spectral efficiency. Meanwhile, other works explore semi-blind OTFS channel estimation frameworks to improve estimation availability and robustness under lower pilot conditions. Although these directions show promise in reducing overhead, ensuring stable convergence and reliable detection simultaneously under DSF, especially fractional Doppler and multipath superposition conditions, remains a key challenge in receiver design.

[0007] Differential Chaotic Keying (DCSK), as a modulation scheme combining spread spectrum and differential correlation detection, has the advantage of low dependence on accurate phase synchronization and accurate channel estimation, thus possessing certain robustness potential in complex channel environments. Combining OTFS and DCSK to form the OTFS-DCSK system is expected to simultaneously leverage the adaptability of OTFS to DSF and the weak coherence (low CSI dependence) advantage of DCSK to improve reliability in high-speed mobile scenarios. Existing research has focused on the detection and performance evaluation of OTFS-DCSK, and proposed schemes to construct low-complexity iterative detectors by utilizing channel sparsity and transmit symbol structure (e.g., rank-1 structure).

[0008] It should be noted that although the above-mentioned OTFS-DCSK detection method utilizes sparsity and structured priors at the algorithm level and adopts iterative frameworks such as the Alternating Direction Multiplier Method (ADMM) to achieve detection, a considerable number of schemes in the published literature still rely on the system assumption that "the receiver can obtain or estimate the equivalent CSI relatively accurately". Therefore, in engineering implementation, control channels or dedicated pilot resources are often still required to support CSI acquisition. At the same time, in order to obtain stable iterative convergence and acceptable bit error rate performance, a trade-off must still be made between pilot / overhead, computational complexity and robustness.

[0009] In summary, existing technologies still have the following shortcomings in OTFS (and OTFS-DCSK) systems under DSF channel conditions: First, the channel state information overhead and rate loss caused by pilots and guard intervals are particularly prominent in high-speed and fractional Doppler scenarios; second, under low pilot or low CSI conditions, the system-level closed-loop scheme for jointly processing channel estimation errors, fractional Doppler spread, and data detection reliability is still insufficient; third, there is still room for improvement in the scheme that integrates the weak CSI dependency of DCSK with the delay-Doppler domain structure advantage of OTFS into a "transceiver structure-detector structure" design, thereby significantly reducing CSI overhead while maintaining achievable complexity and stable convergence. Summary of the Invention

[0010] The purpose of this invention is to address the problems existing in the background art by providing a low channel state information overhead transceiver for dual-select fading channels.

[0011] Therefore, the present invention adopts the following technical solution: A low channel state information overhead transceiver for dual-select fading channels includes a transmitter and a receiver, wherein the transmitter includes: A modulation mapping and serial-to-parallel conversion module, which is used to map the input bitstream to... BPSK Symbols are generated and multi-path parallel symbol sequences are formed. s n , A chaotic sequence generation module, wherein the chaotic sequence generation module can generate a sequence of lengths of... β Reference chaotic sequence x k ; The spread spectrum coupling module is used to implement symbol sequences. s n With reference chaotic sequence x k The elements are coupled element by element to obtain DDThe spread spectrum symbol column vector is connected to the modulation mapping and serial-to-parallel conversion module and the chaotic sequence generation module. A structured frame construction module, the structured frame construction module being used to... DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix is ​​then used to write the data column, anchor column, pilot column, and protection column into the index set. DD Domain emission matrix, forming DD The domain structured frame, index mapping and write control module is connected to the multiplication operation module and the chaotic sequence generation module; The inverse symmetric Fourier transform and Heisenberg transform module is used to perform... DD The domain structured frame performs an inverse sine Fourier transform to map to the TF domain, and then performs a Heisenberg transform on the TF domain symbols to obtain a parallel time-domain transmission signal block. The inverse sine Fourier transform and Heisenberg transform module is connected to the structured frame construction, index mapping and writing control module. The parallel-to-serial conversion module can convert parallel time-domain sample sequences into serial time-domain transmitted signals for output. The parallel-to-serial conversion module is connected to the inverse symptotic Fourier transform and Heisenberg transform module. The receiving end includes: The serial-to-parallel conversion module converts the time-domain sampling sequence obtained by the receiving end according to the time slot index. The sample is divided into blocks and then converted from serial to parallel to obtain a matrix of received samples. The Wigner and SFFT transformation and buffer module is used to sequentially perform Wigner transformation and SFFT transformation on the received samples to obtain the DD domain observation matrix. The Wigner and SFFT transformation and buffer module is connected to the serial-to-parallel conversion module. A weak CSI joint estimation and detection module is used to initialize weak channel state information obtained from a small number of pilots and use the DD domain observation matrix as the joint processing object. In the iterative loop, it sequentially performs multiple rounds of structured detection, constraint projection, and equivalent channel update to output the DD domain structured frame estimate. The weak CSI joint estimation and detection module is connected to the Wigner and SFFT transform and buffer module. A chaotic despreading and parallel-to-serial conversion module, wherein the chaotic despreading and parallel-to-serial conversion module is used to... DDThe reference column estimator and the data column estimator in the domain structured frame estimator undergo chaotic despreading and differential demodulation to obtain the estimated symbol sequence. The estimated symbol sequence is then converted from parallel to serial to output the estimated BPSK bit stream.

[0012] Furthermore, the spread spectrum coupling module includes multiple multipliers, and the chaotic sequence generation module is connected to each multiplier respectively.

[0013] Furthermore, the parallel-to-serial conversion module is connected to a cyclic prefix insertion module, which is used to insert a cyclic prefix before the time-domain transmitted signal to suppress inter-symbol interference.

[0014] Furthermore, the serial-to-parallel conversion module is connected to a cyclic prefix removal module, which is used to remove the cyclic prefix of the received sample.

[0015] A low-channel-state-information-overhead transceiver method for dual-selective fading channels, employing the aforementioned low-channel-state-information-overhead transceiver for dual-selective fading channels, is characterized by comprising a transmission method and a reception method, specifically: The launch method includes the following steps: (1) Map the input bitstream to BPSK Symbols are generated and multi-path parallel symbol sequences are formed. s n ; (2) The symbol sequence s n With length β Reference chaotic sequence x k The elements are coupled element by element to obtain DD Domain spread spectrum symbol column vector; (3) DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix is ​​then used to write the data column, anchor column, pilot column, and protection column into the index set. DD Domain emission matrix, forming DD Domain-structured frames; (4) Perform inverse symplectic Fourier transform and Heisenberg transform on the DD domain structured frame to obtain parallel time-domain transmission signal blocks; (5) Convert the parallel time-domain transmit signal block into a serial time-domain transmit signal output; The receiving method includes the following steps: (1) Convert the received time-domain sampling sequence from serial to parallel to obtain the received samples in matrix form; (2) Perform Wigner transform and SFFT transform on the received samples in sequence to obtain the DD domain observation matrix; (3) Using the weak channel state information obtained from a small number of pilots as the initialization input, and the DD domain observation matrix as the joint processing object, multiple rounds of structured detection, constraint projection and equivalent channel update output are executed in the loop iteration to obtain the DD domain structured frame estimate. (4) Perform chaotic despreading and differential demodulation on the DD domain structured frame estimator to obtain the estimated symbol sequence, and then output the estimated BPSK bit stream through parallel-to-serial conversion.

[0016] Furthermore, in step (5) of the transmission method, a cyclic prefix is ​​inserted before the time-domain transmission signal.

[0017] Furthermore, in step (1) of the receiving method, the received time-domain sampled sequence is converted from serial to parallel after removing the cyclic prefix.

[0018] The beneficial effects of this invention are as follows: (1) In the delay-Doppler (DD) domain frame structure of OTFS-DCSK, this invention introduces only a small number of known auxiliary resources, including anchor columns for locking scale / symbol polarity, sparse pilot columns for initializing and obtaining minimum necessary channel information, and guard columns for suppressing leakage and isolating auxiliary information. When the CSI budget is limited, it can provide the receiver with "minimum necessary identifiable information" to avoid detection drift and performance degradation caused by scale indeterminacy, symbol flipping or insufficient identifiability under weak CSI. At the same time, compared with the high-density pilot structure, the auxiliary resources are occupied less, which is more in line with the low-overhead design goal of engineering implementation.

[0019] (2) The receiver of this invention adopts a joint estimation and detection process of "initialization + closed-loop iteration". First, it obtains initial channel information based on anchoring / pilot and constructs a candidate delay-Doppler parameter set for sparse / quasi-sparse characterization of the DSF channel. By iteratively and alternately executing "channel update ↔ data detection", the detection result feeds back to the channel update, and the channel update feeds back to the detection, thereby gradually reducing the limitation of residual estimation error on detection under low CSI overhead. The structured iterative solution method avoids direct inversion of large-scale matrices, reduces implementation complexity, and improves reproducibility and stability in weak CSI scenarios. Fusion of structural prior and hard constraint projection (rank-1 structural constraint + Return to projection + stop control) (3) In view of the structural characteristics of the OTFS-DCSK transmitted signal, the present invention introduces equivalent structural priors in the detection process and incorporates the known information of the corresponding positions of anchor / pilot / protection zone into the iteration process in the form of hard constraint back projection: after each round of update, the corresponding column / position is back projected to the known value or to satisfy the protection zone constraint, so as to suppress solution space drift and improve convergence reliability.

[0020] Meanwhile, this invention sets up stopping control mechanisms such as residual threshold and iteration upper limit to ensure detection performance while keeping complexity controllable, so that the algorithm has better engineering robustness under weak CSI and model mismatch (such as energy diffusion / quasi-sparseness caused by fractional Doppler). Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the transmitter of the present invention; Figure 2 This is a schematic diagram of the receiver of the present invention; Figure 3 This is a comparison chart of the BER performance of the closed-loop joint estimation and detection mechanisms under weak CSI conditions. Detailed Implementation

[0022] The present invention will now be described in detail with reference to the accompanying drawings and embodiments: A low channel state information overhead transceiver for dual-select fading channels includes a transmitter and a receiver. The transmitter includes: The modulation mapping and serial-to-parallel conversion module is used to map the input bitstream to... BPSK Symbols are generated and a multi-path parallel symbol sequence of length N is formed. s n , n= 0,1,…, N- 1, s n In engineering, the modulation order can also be extended to higher orders. For ease of information structure definition, it is agreed that n=0 corresponds to the reference sequence, and its symbol is fixed as s0=1; the remaining n=1,2,…,N−1 are the symbols corresponding to the information carrying sequence.

[0023] The chaotic sequence generation module can generate sequences of length... β Reference chaotic sequence x k ,in x k , k= 0,1,…, β- 1. Using the second-order Chebyshev polynomial function (CPF) generation method: , 0≤ k≤β- 1. Let the reference chaotic sequence vector be . To ensure consistent spread spectrum energy in each frame, energy normalization can be performed. .

[0024] The spread spectrum coupling module is used to implement symbol sequences. sn With reference chaotic sequence x k Each element is coupled element-wise to obtain the DD domain spread spectrum symbol column vector, which is the BPSK symbol sequence. s n With reference chaotic sequence x k The spread spectrum coupling module consists of multiple multipliers, and in this embodiment, three multipliers are used. For example, for the first multiplier... n Column corresponding symbols The first chaotic sequence with reference k element x k ,have d k,n =s n x k ， Therefore, the first The column vector of the DD domain spreading symbols corresponding to the column can be represented as: , among which, the n Column D :,n It is a column vector of spread spectrum symbols in the DD domain. The frequency coupling module is connected to the modulation mapping and serial-to-parallel conversion module and the chaotic sequence generation module.

[0025] The spread spectrum coupling module is connected to a structured frame construction module, which is used to... DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix D, and , its first n Listed as Let the symbol vector Then it has the form of an outer product. Therefore, in the ideal noise-free case, D is a rank-1 matrix, and this structure can serve as a priori for structured detection at the receiver (i.e., rank-1 / fixed-c). r (Equivalent prior), and then write the data columns, anchor columns, pilot columns, and protection columns according to the index set. DD Domain emission matrix, forming DD Domain-structured frames, specifically: first, define an index set, where: 1) The data column is a data set. Used to carry unknown data symbols; 2) Anchor point set : Carry known Anchor symbols (used to eliminate unrecognizable features such as symbol flipping / scale indeterminacy); 3) Pilot Set : Carries known Pilot symbols (used for weak CSI initialization and parameter candidate set construction); 4) Protection set : Bearer of zero symbol (Guard, used to suppress leakage interference and error propagation).

[0026] The four sets are pairwise disjoint and satisfy the following conditions: ,and This is true by default (the reference sequence can be considered a required Anchor column), and should be written according to the following rules:

[0027] The DD domain structured emission matrix shows the first... The structured symbolic scalar corresponding to the column is used to represent the final value of the column after structured writing. This is obtained after the writing is complete. DD Domain-structured emission matrix ,and and DD Domain structured frames and .

[0028] The structured frame construction module is connected to the inverse symptotic Fourier transform and Heisenberg transform modules, which are used to construct the DD domain structured frame. Perform an inverse symplectic Fourier transform to map to the TF domain and obtain the TF domain symbol. Specifically, follow the formula below:

[0029] in, n Indicates a DD field column index. k Indicates the DD field row index; Indicates the TF domain slot index. Indicates TF domain subcarrier index Then for the TF field symbol Performing the Heisenberg transform yields a parallel time-domain transmitted signal block. Specifically, under the rectangular pulse assumption, the Heisenberg transform can be obtained by indexing the frequency domain. IDFT implementation, and in β When the power of 2 is used, IFFT is used for efficient calculation:

[0030] Depend on Composition β × N A time-domain block can be viewed as a block of parallel time-domain transmitted signals.

[0031] The inverse symmetric Fourier transform and Heisenberg transform modules are connected to a parallel-to-serial conversion module, which can convert the parallel time-domain sample sequence into a serial time-domain transmitted signal output. The parallel-to-serial conversion module is also connected to a cyclic prefix insertion module. At the end of the transmitter, the cyclic prefix insertion module can insert a cyclic prefix (CP) into each parallel time-domain transmitted signal block to suppress ICI / ISI. Finally, the serial conversion module splices the "CP-added parallel time-domain transmitted signal blocks" in time order to form the final time-domain transmitted sequence and transmits it through the DSF channel.

[0032] The receiving end includes: The serial-to-parallel conversion module outputs a time-domain sampled sequence from the transmitter via the DSF channel. The receiver then obtains the time-domain sampled sequence by time slot index. n The sample is divided into blocks and transformed to obtain a parallel matrix of received samples. ,in, i Indicates the sampling index within each time slot (as opposed to the transmitter). In i Consistent) n This represents the time slot index. In addition, the serial-to-parallel conversion module is also connected to a cyclic prefix removal module, which can remove the cyclic prefix (CP removal) of the received samples to suppress inter-block interference and ensure that the subsequent OTFS equivalent model holds true.

[0033] The serial-to-parallel conversion module is connected to a Wigner and SFFT transform and buffer module. The Wigner and SFFT transform and buffer module can sequentially perform Wigner transform and SFFT transform on the received samples to obtain the DD domain observation matrix. Specifically, for the received samples with CP removed, in each time slot n inner edge i Performing the Wigner transform on the dimensionality, under the rectangular impulse assumption, the Wigner transform can be implemented by FFT, yielding time-frequency (TF) domain samples { q k,n}:

[0034] in, k For frequency domain indexing, n For time slot index, i Time-domain sampling index for each time slot, β To represent the number of sampling points corresponding to each time slot, then the time-frequency (TF) domain samples { q k,n Applying SFFT (i.e., performing SFFT transformation) yields the delayed-Doppler (DD) domain observations. w n,k :

[0035] in , N The number of columns in the DD domain structured frame corresponds to the number of parallel symbol sequences at the transmitter, and then the delay-Doppler (DD) domain observations can be processed. w n,k Organized as a DD domain observation matrix ,and This data is then buffered to form the standard input for subsequent detection modules. Furthermore, based on the OTFS equivalent DD field input-output relationship, column-major vectorization can be defined, enabling... Based on the OTFS equivalent channel model, the DD domain observation matrix It can be written as:

[0036] in, This is the DD-domain equivalent channel matrix, which belongs to the idealized real channel, and , For the equivalent noise vector, The vectorized form of the transmitter's DD-domain transmission matrix, and Utilizing the rank-1 structure of the transmitter OTFS-DCSK ,have Therefore, the unknown quantity can be obtained from Nβ Dimensional compression N 3D symbol vector This provides usable priors for structured detection.

[0037] The Wigner module is connected to the SFFT transform and buffer module, and a weak CSI joint estimation and detection module is also connected. This module uses weak channel state information obtained from a small number of pilots as initialization input, and the DD domain observation matrix is ​​used as the initialization input. This provides usable priors for structured detection.

[0038] The Wigner module is connected to the SFFT transform and buffer module, and a weak CSI joint estimation and detection module is also connected. This module uses weak channel state information obtained from a small number of pilots as initialization input, and the DD domain observation matrix is ​​used as the initialization input. The loop iteration stops when the result changes below a preset threshold. After the loop iteration stops, the final output is the equivalent channel estimation matrix. and DD domain structured frame estimator ,in, This represents the final estimated equivalent channel matrix, used for round-by-round updates of the channel representation during the closed-loop iteration process. This represents the final output DD domain structured frame estimate, which serves as the direct input to the subsequent chaotic despreading decision and output recovery module.

[0039] Specifically, under weak CSI conditions, the receiver does not have a complete equivalent channel representation in advance and can only obtain initialization information by relying on the known structural information written by the transmitter. Therefore, the set of known structural information written by the transmitter is utilized. Construct an initial channel representation; where, , and Let Anchor, Pilot, and Guard column indices represent the sets of column indices, respectively. Under the condition that the dual-selection fading (DSF) channel exhibits sparse or quasi-sparse characteristics, a candidate delay-Doppler set is further introduced. ,in, Indicates the first The delay parameters corresponding to each candidate path Indicates the first Doppler parameters corresponding to each candidate path For candidate path index, L The candidate delay-Doppler set represents the total number of candidate delay-Doppler pairs. It can be used to generate initial candidate supports based on the observation information and candidate set corresponding to anchor points / pilots. An initial equivalent channel estimation matrix can be constructed. As the initial input for alternating equivalent channel updates and structured detection in the iterative loop, in this embodiment, during the iterative loop, when the loop reaches the t-th round, it first calculates the equivalent channel estimation matrix... Given the conditions, the observation matrix in the DD domain Structured detection is performed. In this embodiment, structured detection can be achieved through an ADMM-based structured detection solution to recover a temporary estimate of the DD domain structured frame for the current round. Specifically, let the vectorized variable to be detected be... ,and ,in, This indicates the number of columns in the DD field structured frame. Let the length of the chaotic sequence or the spreading length corresponding to each column be denoted as the optimization variable. express The estimate was obtained, and auxiliary variables were introduced. , used to characterize low-dimensional solutions that satisfy structural constraints, can be made , to enforce order (or fixed-c) r The structure prior is given by Vec(·), which denotes the vectorization operator. Structure detection can be performed according to the following formula:

[0040] After obtaining the structured detection results for the current round, the projection operation can be constrained, applying Anchor / Pilot / Guard constraints as well as rank-1 / fixed-c constraints. r The structure prior is projected back onto the feasible solution domain to suppress the solution space drift caused by incomplete channel knowledge under weak CSI conditions, and a DD-domain structured frame estimator that satisfies the structural constraints is obtained. , The corresponding DD-domain structured frame recovery result under the current outer loop is then used to estimate the DD-domain structured frame based on the constrained projection. The matching relationship between the DD domain observation matrix and the equivalent channel representation is used to update the equivalent channel estimation matrix required for the next round of iteration. Once the cyclic iteration stops, the DD domain structured frame estimate can be obtained. DD domain structured frame estimator It can be used as a direct input for subsequent chaos despreading decision and output recovery modules.

[0041] The weak CSI joint estimation and detection module is connected to a chaotic despreading and parallel-to-serial conversion module, which is used to estimate the structured frame in the DD domain. Perform chaotic despreading and differential demodulation to obtain the estimated symbol sequence, and then perform parallel-to-serial conversion on the estimated symbol sequence to output the estimated BPSK bit stream.

[0042] This invention also provides a low channel state information overhead transceiver method for dual-select fading channels, employing the aforementioned low channel state information overhead transceiver for dual-select fading channels, including a transmission method and a reception method, specifically: The launch method includes the following steps: (1) Map the input bitstream to BPSK Symbols are generated and a multi-path parallel symbol sequence is formed.

[0043] (2) The symbol sequence is coupled element by element with the reference chaotic sequence to obtain DD Domain spread spectrum symbol column vector.

[0044] (3) DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix is ​​then used to write the data column, anchor column, pilot column, and protection column into the index set. DD Domain emission matrix, forming DD Domain-structured frames.

[0045] (4) Perform inverse symptotic Fourier transform and Heisenberg transform on the DD domain structured frame to obtain parallel time-domain transmission signal blocks.

[0046] (5) Insert a cyclic prefix (CP) into each parallel time-domain transmit signal block to suppress ICI / ISI. Finally, the serial conversion module splices the parallel time-domain transmit signal blocks with CP in time order and converts them into serial time-domain transmit signal output.

[0047] The receiving method includes the following steps: (1) Remove the cyclic prefix (CP) from the received time-domain sampled sequence and then perform serial-to-parallel conversion to obtain the received sample in matrix form.

[0048] (2) The Wigner transform and SFFT transform are performed on the received samples in sequence to obtain the DD domain observation matrix.

[0049] (3) Using the weak channel state information obtained from a small number of pilots as the initial input, and the DD domain observation matrix as the joint processing object, multiple rounds of structured detection, constraint projection and equivalent channel update output are executed in the loop iteration to obtain the DD domain structured frame estimate; (4) Perform chaotic despreading and differential demodulation on the DD domain structured frame estimator to obtain the estimated symbol sequence and output the estimated BPSK bit stream after parallel-to-serial conversion.

[0050] To verify the effectiveness of this invention in detecting low channel state information overhead under dual-selective fading channels, four receiver conditions were constructed for comparative testing using the same OTFS-DCSK system parameters and channel conditions: the first is the perfect CSI case, which serves as the upper bound of system performance; the second is the high CSI overhead case, which obtains more sufficient channel information by increasing pilot / training resources; the third is the low CSI overhead closed-loop joint estimation and detection scheme proposed in this invention (Anchor / Pilot / Guard minimum training information structure + weak CSI initialization + closed-loop update + constrained back projection); and the fourth is a control scheme with low CSI overhead but no closed loop.

[0051] Experimental results show that, Figure 3As shown, under perfect CSI, the system BER decreases monotonically with increasing SNR, constituting an achievable upper bound for performance. Under high CSI overhead, due to sufficient training information, the BER curve remains highly consistent with the perfect CSI upper bound, indicating that the detection framework and noise calibration are correct and can approach optimal performance under sufficient CSI conditions. In contrast, when only a one-time weak CSI initialization is used without loop closure, the rate of decrease in bit error rate slows significantly with the increase of Eb / N0, reflecting that estimation error becomes the dominant factor in performance when training information is limited. Furthermore, the present invention, while maintaining low CSI overhead, iteratively corrects the channel information through closed-loop joint estimation and detection and constrained back projection mechanism, which can continuously reduce BER in the medium-to-high SNR region and significantly suppress the bit error rate saturation trend that appears in the control scheme, so that the overall BER curve is located between the "high CSI overhead scheme" and the "low CSI no-loop scheme" and is closer to the perfect CSI upper bound.

[0052] Therefore, it can be seen that the Anchor / Pilot / Guard minimum training information structure and closed-loop joint estimation-detection mechanism adopted in this invention can achieve more reliable bit error rate performance with lower CSI acquisition overhead under dual-selection fading channel conditions, verifying the technical effect of this invention on "stable detection can still be achieved with low CSI overhead".

Claims

1. A low channel state information overhead transceiver for dual-select fading channels, characterized in that, It includes a transmitter and a receiver, wherein the transmitter includes: A modulation mapping and serial-to-parallel conversion module, which is used to map the input bitstream to... BPSK Symbols are generated and multi-path parallel symbol sequences are formed. s n , A chaotic sequence generation module, wherein the chaotic sequence generation module can generate a sequence of lengths of... β Reference chaotic sequence x k ; The spread spectrum coupling module is used to implement symbol sequences. s n With reference chaotic sequence x k The elements are coupled element by element to obtain DD The spread spectrum symbol column vector is connected to the modulation mapping and serial-to-parallel conversion module and the chaotic sequence generation module. A structured frame construction module, the structured frame construction module being used to... DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix is ​​then used to write the data column, anchor column, pilot column, and protection column into the index set. DD Domain emission matrix, forming DD The domain structured frame, index mapping and write control module is connected to the multiplication operation module and the chaotic sequence generation module; The inverse symmetric Fourier transform and Heisenberg transform module is used to perform... DD The domain structured frame performs an inverse sine Fourier transform to map to the TF domain, and then performs a Heisenberg transform on the TF domain symbols to obtain a parallel time-domain transmission signal block. The inverse sine Fourier transform and Heisenberg transform module is connected to the structured frame construction, index mapping and writing control module. The parallel-to-serial conversion module can convert parallel time-domain sample sequences into serial time-domain transmitted signals for output. The parallel-to-serial conversion module is connected to the inverse symptotic Fourier transform and Heisenberg transform module. The receiving end includes: The serial-to-parallel conversion module converts the time-domain sampling sequence obtained by the receiving end according to the time slot index. The sample is divided into blocks and then converted from serial to parallel to obtain a matrix of received samples. The Wigner and SFFT transformation and buffer module is used to sequentially perform Wigner transformation and SFFT transformation on the received samples to obtain the DD domain observation matrix. The Wigner and SFFT transformation and buffer module is connected to the serial-to-parallel conversion module. A weak CSI joint estimation and detection module is used to initialize weak channel state information obtained from a small number of pilots and use the DD domain observation matrix as the joint processing object. In the iterative loop, it sequentially performs multiple rounds of structured detection, constraint projection, and equivalent channel update to output the DD domain structured frame estimate. The weak CSI joint estimation and detection module is connected to the Wigner and SFFT transform and buffer module. A chaotic despreading and parallel-to-serial conversion module, wherein the chaotic despreading and parallel-to-serial conversion module is used to... DD The reference column estimator and the data column estimator in the domain structured frame estimator undergo chaotic despreading and differential demodulation to obtain the estimated symbol sequence. The estimated symbol sequence is then converted from parallel to serial to output the estimated BPSK bit stream.

2. A low channel state information overhead transceiver for dual-select fading channels according to claim 1, characterized in that, The spread spectrum coupling module includes multiple multipliers, and the chaotic sequence generation module is connected to each multiplier respectively.

3. A low channel state information overhead transceiver for dual-select fading channels according to claim 1, characterized in that, The parallel-to-serial conversion module is connected to a cyclic prefix insertion module, which is used to insert a cyclic prefix before the time-domain transmitted signal to suppress inter-symbol interference.

4. A low channel state information overhead transceiver for dual-select fading channels according to claim 1, characterized in that, The serial-to-parallel conversion module is connected to a cyclic prefix removal module, which is used to remove the cyclic prefix from the received samples.

5. A low channel state information overhead transceiver method for dual-select fading channels, employing the low channel state information overhead transceiver for dual-select fading channels as described in any one of claims 1-4, characterized in that, This includes transmission and reception methods, specifically: The launch method includes the following steps: (1) Map the input bitstream to BPSK Symbols are generated and multi-path parallel symbol sequences are formed. s n ; (2) The symbol sequence s n With length β Reference chaotic sequence x k The elements are coupled element by element to obtain DD Domain spread spectrum symbol column vector; (3) DD The domain spread spectrum symbol column vector is organized as follows DD The domain emission matrix is ​​then used to write the data column, anchor column, pilot column, and protection column into the index set. DD Domain emission matrix, forming DD Domain-structured frames; (4) DD The domain structured frame is subjected to inverse symptotic Fourier transform and Heisenberg transform to obtain parallel time-domain transmitted signal blocks; (5) Convert the parallel time-domain transmit signal block into a serial time-domain transmit signal output; The receiving method includes the following steps: (1) Convert the received time-domain sampling sequence from serial to parallel to obtain the received samples in matrix form; (2) Perform Wigner transform and SFFT transform on the received samples in sequence to obtain the DD domain observation matrix; (3) Using the weak channel state information obtained from a small number of pilots as the initialization input, and with DD The domain observation matrix is ​​used as a joint processing object. In the cyclic iteration, multiple rounds of structured detection, constraint projection and equivalent channel update are performed sequentially to obtain the DD domain structured frame estimate. (4) Perform chaotic despreading and differential demodulation on the DD domain structured frame estimator to obtain the estimated symbol sequence, and then output the estimated BPSK bit stream through parallel-to-serial conversion.

6. A low channel state information overhead transceiver method for dual-select fading channels according to claim 5, characterized in that, In step (5) of the transmission method, a cyclic prefix is ​​inserted before the time-domain transmission signal.

7. A low channel state information overhead transceiver method for dual-select fading channels according to claim 5, characterized in that, In step (1) of the receiving method, the received time-domain sampled sequence is converted from serial to parallel after removing the cyclic prefix.

Images