A single-carrier underwater acoustic communication method and device under deep-sea long-delay clustered channel.

Abstract

This invention discloses a single-carrier underwater acoustic communication method and apparatus for deep-sea long-delay clustered channels. The method considers a single-carrier block transmission system and, based on a single-carrier phase-shift keying modulation scheme, modulates communication information onto the carrier phase. The receiver decoding process is as follows: First, channel structure detection is performed, calculating the block length spanned by the long-delay channel and the margin for any insufficient blocks. Then, initial interference cancellation and equalization are performed to obtain initial symbol values. Next, a joint posterior density function of the channel and symbols is constructed using a joint channel estimation, interference cancellation, and symbol detection method. The channel is jointly estimated through message propagation at each side, interference is reconstructed and cancelled, and finally, symbols are detected. This invention effectively eliminates inter-block interference, improves the robustness of single-carrier block transmission systems in deep-sea long-delay clustered channels, and can obtain relatively accurate channel values ​​even with insufficient training sequences, thereby reducing interference cancellation errors. Furthermore, it has acceptable computational complexity.

Description

Technical Field

[0001] This invention relates to the field of deep-sea acoustic communication, and more specifically to a single-carrier underwater acoustic communication method and apparatus under a deep-sea long-delay clustered channel. Background Technology

[0002] In underwater acoustic single-carrier block transmission systems, the guard interval is typically set to be greater than the channel's maximum multipath delay to avoid inter-block interference. However, in some deep-sea scenarios, the channel exhibits long-delay spread and sparse clustering characteristics, with the total channel length often reaching tens or even hundreds of symbols, far exceeding the guard interval length. This leads to severe inter-block interference, and relying solely on the guard interval to prevent inter-block interference in such cases significantly reduces the system's transmission rate. Existing inter-block interference suppression methods for long-delay spread channels are mostly designed for OFDM systems, while in single-carrier block transmission systems, inter-block interference elimination is largely limited to short-delay spread channels and is no longer applicable to the current scenario.

[0003] Excessive channel delay can lead to insufficient training data for channel estimation, resulting in inaccurate channel estimation results, inadequate interference cancellation, the introduction of additional errors, and a reduction in the performance of the underwater acoustic communication system. Therefore, there is an urgent need to design a single-carrier underwater acoustic communication method that can maintain robust performance even with limited observation data under deep-sea long-delay clustered channels. Summary of the Invention

[0004] Purpose of the invention: In order to improve the robustness of underwater acoustic single-carrier communication systems under deep-sea long-delay clustered channels, this invention provides a single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels. This method improves the robustness of single-carrier communication under deep-sea long-delay clustered channels while having acceptable computational complexity.

[0005] Another objective of this invention is to provide a single-carrier underwater acoustic communication device for deep-sea long-delay clustered channels.

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

[0007] A single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels is applied at the receiver. The method consists of two parts: the first part obtains the initial value of the transmitted symbol through preprocessing; the second part obtains the detected symbol through joint channel estimation, interference cancellation, and symbol detection. The first part includes steps 1-6:

[0008] Step 1: Receive the passband signal The baseband signal is obtained after demodulation and downsampling. n represents the nth block of received signal, and N represents the total number of received signal blocks;

[0009] Step 2: Based on the passband signal from Step 1 The signal is matched and correlated with the hyperbolic FM signal used as a synchronization signal during transmission, and the output signal is the matched and correlated signal.

[0010] Step 3: Based on the matched correlation signal output in Step 2 The specific structure of the channel is determined by detecting the position of the synchronization peak, and the channel structure parameters are output. Δτ represents the channel spacing, ζ n This indicates the channel length margin that is not a complete block.

[0011] Step 4: Training based on the baseband signal output in Step 1 and the channel structure parameters output in step 3 Determine the block used for channel estimation, the length of the corresponding training sequence, and the length of the estimated channel. Output the estimated channel through which the nth block symbol passes.

[0012] Step 5: Based on the baseband received signal output in Step 1 and the channel output in step 4 Interference cancellation is performed to obtain the signal after interference cancellation.

[0013] Step 6: Based on the channel output in Step 4 The signal after interference cancellation output in step 5 After equalization processing, the detected symbols are obtained.

[0014] The first part processes the received signal block by block using steps 1 to 6, and outputs the first to Nth blocks of received signals. Channel structure and the symbols after detection

[0015] Part Two includes steps 7-10:

[0016] Step 7: Receive signals from block 1 to block N based on the first part of the input. Channel structure and the symbols after detection Simultaneously, based on the detection symbols output in step 9 The noise variance output in step 10 With hyperparameter γ, a Gaussian generalized approximate message-passing method based on damping factor modulation is used for channel estimation in the first iteration. The initial value is the symbol obtained from the first part of the detection. The initial values ​​for noise variance and hyperparameters are set to 1, and the output iterations are performed up to the maximum number of times. Time-estimated channel

[0017] Step 8: Based on the channel estimated in Step 7 and the first part of the input signal Perform interference reconstruction and elimination, and output the signal after eliminating the previous interference.

[0018] Step 9: Based on the channel estimated in Step 7 Step 8: Signal after interference removal and the signal output from the first part Simultaneously, combine the noise variance output from step 10. The hyperparameter γ is used to detect the current block symbol. The initial value is set in the same way as in step 7;

[0019] Step 10: Based on the channel output in Step 7 The symbol output in step 9 and the signal output from the first part Update noise variance Along with the hyperparameter γ, the updated value is returned to steps 7 and 9 to iterate again until the iteration condition is met or the iteration number is reached. This concludes the second part, and the detected symbol is output.

[0020] This invention also provides a single-carrier underwater acoustic communication device for deep-sea long-delay clustered channels, applied at the receiving end, comprising: a first processing unit for obtaining initial values ​​of transmitted symbols through preprocessing, and a second processing unit for obtaining detected symbols through joint channel estimation, interference cancellation, and symbol detection, wherein the first processing unit includes:

[0021] Signal demodulation module: for receiving passband signals The baseband signal is obtained after demodulation and downsampling. n represents the nth block of received signal, and N represents the total number of received signal blocks;

[0022] Signal matching module: based on the passband signal of the input demodulation module. The signal is matched and correlated with the hyperbolic FM signal used as a synchronization signal during transmission, and the output signal is the matched and correlated signal.

[0023] Channel structure parameter determination module: based on the matched correlation signal output by the signal matching module. The specific structure of the channel is determined by detecting the position of the synchronization peak, and the channel structure parameters are output. Δτ represents the channel spacing, ζ n This indicates the channel length margin that is not a complete block.

[0024] First channel estimation module: based on the training portion of the baseband signal output from the signal demodulation module. Channel structure parameters output by the channel structure parameter determination module Determine the block used for channel estimation, the length of the corresponding training sequence, and the length of the estimated channel. Output the estimated channel through which the nth block symbol passes.

[0025] First interference cancellation module: Based on the baseband received signal output by the signal demodulation module. and the channel output by the channel estimation module Interference cancellation is performed to obtain the signal after interference cancellation.

[0026] First symbol detection calculation module: based on the channel output of the channel estimation module. The signal after interference cancellation output by the interference cancellation module After equalization processing, the detected symbols are obtained.

[0027] The first processing unit processes the received signal block by block from the signal demodulation module to the detection symbol calculation module, and outputs the received signals from the first block to the Nth block. Channel structure and the symbols after detection To the second processing department;

[0028] The second processing unit includes:

[0029] Second channel estimation module: Based on the received signals from the first to the Nth blocks input by the first processing unit. Channel structure and the symbols after detection Simultaneously, based on the detection symbols output by the second detection symbol calculation module noise variance of the process control module output With hyperparameter γ, a Gaussian generalized approximate message-passing method based on damping factor modulation is used for channel estimation in the first iteration. The initial value is the symbol obtained from the first part of the detection. The initial values ​​for noise variance and hyperparameters are set to 1, and the output iterations are performed up to the maximum number of times. Time-estimated channel

[0030] Second interference cancellation module: Based on the channel estimated by the second channel estimation module... and the signal output by the first processing unit Perform interference reconstruction and elimination, and output the signal after eliminating the previous interference.

[0031] Second symbol detection calculation module: Based on the channel estimated by the second channel estimation module... The second interference cancellation module eliminates the interference from the signal. and the signal output by the first processing unit Simultaneously, the noise variance output by the process control module is considered. With hyperparameter γ, used to detect the current block symbol The initial values ​​are set the same as those in the second channel estimation module;

[0032] Process control module: Based on the channel output of the second channel estimation module The symbols output by the second detection symbol calculation module and the signal output by the first processing unit Update noise variance Along with the hyperparameter γ, the updated value is returned to the second channel estimation module and the second symbol detection calculation module to iterate again until the iteration condition is met or the iteration number is reached. At this point, the operation of the second processing unit ends, and the detected symbol is output.

[0033] The present invention also provides a computing processing device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, wherein when the programs are executed by the processors, they implement the steps of the single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels as described above.

[0034] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels as described above.

[0035] Beneficial Effects: This invention, based on a single-carrier system, obtains initial symbol values ​​by initially eliminating interference, and then reconstructs and eliminates the interference by combining channel estimation, interference cancellation, and symbol detection methods, outputting the final estimated symbol. Compared with existing technologies, this invention has the following beneficial effects:

[0036] (1) Improving the reliability of single-carrier communication under long-delay clustered channels. Communication reliability is highly correlated with the performance of data processing methods. Under long-delay clustered channels, conventional time-domain equalizers require long equalizer lengths, leading to difficulty in convergence and high complexity, resulting in a severe decrease in communication reliability; and due to the existence of inter-block interference, the conditions for frequency-domain equalization are not met, making it impossible to perform. This invention adopts a method of joint channel estimation, interference cancellation, and symbol detection, which effectively eliminates inter-block interference caused by channel delay and improves communication reliability.

[0037] (2) High-performance decoding of single-carrier communication data can also be achieved when there is insufficient training sequence. Excessive channel delay may lead to insufficient training sequence, resulting in less observation data for channel estimation, poor channel estimation results, insufficient interference cancellation, or even the introduction of new errors. The method of combining channel estimation, interference cancellation, and symbol detection in this invention uses all symbols for channel estimation, making the results more accurate and effectively realizing high-performance decoding of single-carrier communication data. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments are briefly introduced below. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort, wherein:

[0039] Figure 1 This is a flowchart of a single-carrier underwater acoustic communication method under a deep-sea long-delay clustered channel;

[0040] Figure 2 This is a schematic diagram of long-delay clustered channels and signal block length;

[0041] Figure 3(a) shows the variation of the equalization bit error rate with the signal-to-noise ratio when the training sequence is 96 under BPSK modulation.

[0042] Figure 3(b) shows the variation of the equalization bit error rate with the signal-to-noise ratio under BPSK modulation when the training sequence is 110.

[0043] Figure 3(c) shows the variation of the equalization bit error rate with the signal-to-noise ratio when the training sequence is 128 under BPSK modulation. Detailed Implementation

[0044] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to specific embodiments and accompanying drawings.

[0045] This invention proposes a single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels. It is a single-carrier communication method that performs signal processing in the time domain. To suppress inter-block interference caused by long-delay channels in single-carrier block transmission, the method first performs interference cancellation and equalization to obtain initial symbol values. Then, it uses a joint channel estimation, interference cancellation, and symbol detection method to reconstruct and cancel the interference, effectively suppressing inter-block interference caused by long-delay clustered channels. At the same time, it can also perfectly suppress interference in scenarios with insufficient training sequences, improve the robustness of communication, and has feasible complexity. It can achieve high-performance decoding of single-carrier communication data under long-delay clustered channels with inter-block interference.

[0046] This invention provides a single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels. The workflow of the method is as follows: Figure 1 Without loss of generality, this invention focuses on the case of arrival through two clusters of channels. The method consists of two parts: the first part obtains the initial values ​​of the transmitted symbols block by block, and the second part further improves performance through a joint channel estimation, interference cancellation, and symbol detection method.

[0047] The first part adopts a block-by-block approach, starting with the first block of received signal and ending with the Nth block. When processing the (n+1)th block of received signal, the detected symbol of the nth block is used. Steps 1 to 6 together constitute the first part. Each step uses the nth block signal as an example, and the subscript outside the curly braces represents the range of values ​​for n. A detailed explanation follows:

[0048] Step 1: Passband signal The signal is converted to baseband through demodulation, downsampling, and low-pass filtering. The baseband received signal is processed in each block. Including signals from the training part All signals that transmit information are output to subsequent steps.

[0049] Step 2: Based on passband signal The signal is matched and correlated with the transmitted hyperbolic frequency modulated signal used for synchronization to obtain the matched and correlated signal.

[0050] Step 3: Based on the matched correlation signal output in Step 2 The location of the peak is detected to determine the time interval between the arrival at the two cluster channels and the length of each cluster channel. This is achieved by considering the interval Δτ between the two cluster channels and the block length T. b Based on the relationship, the number of blocks traversed by the channel is calculated. and channel length margin less than one block length Δτ=Δ n T b +ζ n This embodiment only considers the case of crossing a single block, i.e., Δ. n =1.

[0051] Step 4: Based on the training portion of the baseband received signal output in Step 1 and the channel structure parameters output in step 3 The lengths of the training sequence and the observation data used for channel estimation are determined, and the channel through which the nth symbol passes is estimated using a sparse Bayesian learning method. And output it.

[0052] Step 5: Based on the baseband received signal output in Step 1 The channel output from step 4 Compared with the n-Δth already detected n block symbols Jointly reconstruct inter-block interference, and from After interference is eliminated, the baseband received signal after interference cancellation for the current block is obtained.

[0053] Step 6: Based on the interference-cancelled baseband received signal output in Step 5 The estimated channel is output from step 4. The symbols of the current block are obtained by equalization using a linear least mean square error algorithm. Then it is input into step 5 of the next symbol detection to eliminate interference with the next received signal.

[0054] Once all block symbols have been equalized, the first part ends. The baseband received signals from all blocks in step 1 are then transferred. Channel structure parameters of all blocks obtained from peak detection in step 3 and the symbols of all blocks obtained from the equalization in step 6. The output is fed into the second part, joint channel estimation, interference cancellation, and symbol detection.

[0055] The second part is also performed block by block, starting with the received signal from the first block and ending with the Nth block. Taking the symbol detection of the nth block as an example, the difference from the first part is that the second part incorporates the (n-Δ)th block... n 、n+Δ n Block (Δ here) n =1) Receive the signal to obtain the nth symbol. Steps 7-10 are part of a joint estimation and detection process, in which the channel and symbols are modeled as a joint posterior distribution. Taking the nth signal as an example, the joint posterior distribution is:

[0056]

[0057] Among them, channel for The vectors formed, L1 and L2 are respectively and The dimension, γ n It is a channel hyperparameters, γ n,l Represent and elements in γ. M and N b These are the baseband received symbols y n and the symbol to be estimated x n dimensionality, y n,m and x n,kEach element represents its respective element. The aforementioned posterior distribution can be represented by a factor graph, and the estimated values ​​of the channel and symbols are calculated through the mutual transmission of messages between the edges. The second part contains a calculation order, which together constitutes the outer iterative process. The maximum number of iterations is U, which controls the overall iteration of steps 7-10. Each step takes the nth signal as an example, and the subscript outside the curly braces represents the range of values ​​for n.

[0058] In this embodiment of the invention, the second part combines the received signals from the (n-1)th and (n+1)th blocks to obtain the nth symbol, and the outer iteration number of the second part is set to U = 15. A detailed explanation follows:

[0059] Step 7: Initialize the symbol based on the output of the first part Channel structure parameters and baseband received symbols and the noise variance output in step 10 And the hyperparameter γ, during the first iteration The initial values ​​of both γ and t are set to 1. The channel is updated through the transmission of messages from each side. The maximum number of iterations in this step is [number missing]. In the t-th iteration, the channel update process for the nth symbol is as follows.

[0060] First, update the mean of the noiseless linear output variable. and variance

[0061]

[0062]

[0063] Where X n-1,m,l , and X n+1,m,l They are matrix X n-1 , and X n+1 The element in row m and column l, X n-1 , They are respectively composed of the symbol x n-1 x n The resulting M×L2 and M×L1 dimensional Toplitz matrices are constructed. X n+1 They are respectively composed of the symbol x n x n+1 The resulting M×L2 and M×L1 dimensional Toplitz matrices. and These are the posterior variance and mean of the channel to be estimated. In the first iteration, the initial values ​​are set to 1 and 0, respectively. It is an intermediate variable; its initial value is set to 0 when t = 1. (Intermediate variable) The inverse residual variance is updated as follows:

[0064]

[0065] in This is the damping factor, set to 0.4 here, which controls the update speed and convergence of the variables.

[0066] Then calculate the input variables. Variables affected by noise and its reciprocal variance

[0067]

[0068] After obtaining the mean and variance, the input variables are recovered from them. l = 1, ..., L, l' = l + L, their values ​​are respectively:

[0069]

[0070] At this point, the corresponding variance is Where γ l It is the l-th element in γ. When the maximum number of iterations is reached... At that time, step 7 was updated.

[0071] Step 8: When the estimate obtained in step 7 is... Using the obtained channel and detected symbols (This is the text) Reconstruct and eliminate the interference in the nth block of the received signal to obtain the nth block of the received signal after interference cancellation.

[0072] Step 9: Receive the nth block of signal after interference elimination in Step 8. And the (n+Δ)th part of the input n Block receive signal (This is the text) The channel estimated in step 7) (This is the text) Combine the (n+1)th received signal to perform symbol detection, and simultaneously incorporate the noise variance output from step 10. The hyperparameter γ is used to detect the current block symbol. The initial value is set in the same way as in step 7.

[0073] In this step, the process is also iterative, and the update method is similar to that in step 7, with the maximum number of iterations set to [value missing]. In the i-th iteration The update is as follows:

[0074] First update the variables. and its variance

[0075]

[0076]

[0077] in, and They are matrices and The element in the m-th row and k-th column, and They are and The M×N structure b Wittoripliz matrix. and These are the posterior variance and mean of the symbol to be detected. In the first iteration, the initial values ​​are set to 1 and 0, respectively. It is an intermediate variable; when i = 1, its initial value is set to 0. (Intermediate variable) The inverse residual variance is updated as follows:

[0078]

[0079] Then calculate the input variables. Variables affected by noise and its reciprocal variance

[0080]

[0081] After obtaining the mean and variance, the input variables are recovered from them. k = 1, ..., N b , k'=k+N b Their values ​​are as follows:

[0082]

[0083]

[0084] At this point, the corresponding variance is Where p k It is the k-th element in the prior variance of the symbol to be estimated. When the maximum number of iterations is reached... At that time, step 9 was updated.

[0085] Step 10: Based on the received signal output from the first part The channel output in step 7 and the symbols output in step 9 The noise variance and hyperparameters of the (u+1)th outer iteration are jointly updated. In the nth symbol block, the update method for both is as follows:

[0086]

[0087]

[0088] The result is then input back into steps 7 and 9 for iteration. When the iteration reaches the maximum number of outer iterations U or the loop termination condition is met, the iteration exits, the second part is completed, and the detected symbol is output.

[0089] The advantages of this invention will be further explained below with reference to simulation results. Obviously, the described simulation results are only some embodiments of this invention, not all embodiments, and are used only for illustration and explanation, not to limit the application of this invention. All other embodiments obtained by those skilled in the art based on the experimental data processing results of this invention without inventive effort should fall within the scope of protection of this invention.

[0090] Experimental conditions: Simulations were performed using the method proposed in this invention to verify its performance. A schematic diagram of the transmitted data and channel structure is shown below. Figure 2 As shown, from Figure 2 As can be seen, the channel exhibits significant long-delay clustering characteristics, with each preceding block causing inter-block interference to the following block. The communication parameters in the simulation are as follows: The simulation uses a single-carrier block transmission system with BPSK modulation. The center frequency, sampling frequency, symbol duration, and roll-off factor of the root-raised cosine filter are set to 2kHz, 20kHz, 1ms, and 1, respectively. Each signal block length is set to 560ms, including a 96ms guard interval and 464ms of data symbols, with five signal blocks per frame.

[0091] Experimental Results Analysis: Figures 3(a), 3(b), and 3(c) show the relationship between bit error rate and signal-to-noise ratio (SNR) when the training sequence lengths are 96, 110, and 128, respectively, with the SNR ranging from -5 to 5 dB. The methods used for comparison are linear minimum mean square error without interference cancellation, linear minimum mean square error with interference cancellation, joint detection with interference cancellation, linear minimum mean square error with interference cancellation (twice), and the invented method.

[0092] As can be seen from Figures 3(a) and 3(b), insufficient training sequences result in insufficient observation data for channel estimation. In this case, the estimated second cluster channel exhibits significant errors. Abnormal channels cause burst errors in several blocks of a certain frame during interference cancellation, which the decoder cannot correct. Consequently, the linear minimum mean square error (LMSE) method for interference cancellation and the joint detection method for interference cancellation are inferior to the method without interference cancellation. The LMSSE method (two iterations) for interference cancellation, due to its use of all symbols for channel estimation and subsequent interference cancellation and equalization, significantly outperforms the other three methods. The invented method continuously updates the channel, noise variance, and symbols in each iteration, constantly correcting the channel estimation results, resulting in the lowest NMSE and lowest bit error rate for the second cluster channel. As can be seen from Figure 3(c), when the training sequence is 128, due to the longer training sequence length, the channel estimation is accurate and the interference cancellation error is small, and the performance of all interference cancellation methods is improved. At this time, the joint detection method for interference cancellation uses all received block information including the current block to recover the symbol, and its performance is even better than the linear minimum mean square error (twice) of line interference cancellation. However, the channel estimation result of the invented method is accurate, and the bit error rate performance is still the best.

[0093] This invention relates to an underwater acoustic communication method with robust communication performance and acceptable computational complexity under deep-sea long-delay clustered channels. The advantages are: (1) effectively eliminating inter-block interference and improving the communication robustness of single-carrier block transmission systems under deep-sea long-delay clustered channels; (2) even when the training sequence is insufficient, relatively accurate channel values ​​can be obtained, thereby reducing interference cancellation errors; (3) while ensuring performance, the use of message passing algorithms has acceptable computational complexity.

[0094] This invention also provides a single-carrier underwater acoustic communication device for deep-sea long-delay clustered channels, applied at the receiving end, comprising: a first processing unit for obtaining initial values ​​of transmitted symbols through preprocessing, and a second processing unit for obtaining detected symbols through joint channel estimation, interference cancellation, and symbol detection, wherein the first processing unit includes:

[0095] Signal demodulation module: for receiving passband signals The baseband signal is obtained after demodulation and downsampling. n represents the nth block of received signal, and N represents the total number of received signal blocks;

[0096] Signal matching module: based on the passband signal of the input demodulation module. The signal is matched and correlated with the hyperbolic FM signal used as a synchronization signal during transmission, and the output signal is the matched and correlated signal.

[0097] Channel structure parameter determination module: based on the matched correlation signal output by the signal matching module. The specific structure of the channel is determined by detecting the position of the synchronization peak, and the channel structure parameters are output. Δτ represents the channel spacing, ζ n This indicates the channel length margin that is not a complete block.

[0098] First channel estimation module: based on the training portion of the baseband signal output from the signal demodulation module. Channel structure parameters output by the channel structure parameter determination module Determine the block used for channel estimation, the length of the corresponding training sequence, and the length of the estimated channel. Output the estimated channel through which the nth block symbol passes.

[0099] First interference cancellation module: Based on the baseband received signal output by the signal demodulation module. and the channel output by the channel estimation module Interference cancellation is performed to obtain the signal after interference cancellation.

[0100] First symbol detection calculation module: based on the channel output of the channel estimation module. The signal after interference cancellation output by the interference cancellation module After equalization processing, the detected symbols are obtained.

[0101] The first processing unit processes the received signal block by block from the signal demodulation module to the detection symbol calculation module, and outputs the received signals from the first block to the Nth block. Channel structure and the symbols after detection To the second processing department;

[0102] The second processing unit includes:

[0103] Second channel estimation module: Based on the received signals from the first to the Nth blocks input by the first processing unit. Channel structure and the symbols after detection Simultaneously, based on the detection symbols output by the second detection symbol calculation module noise variance of the process control module output With hyperparameter γ, a Gaussian generalized approximate message-passing method based on damping factor modulation is used for channel estimation in the first iteration. The initial value is the symbol obtained from the first part of the detection. The initial values ​​for noise variance and hyperparameters are set to 1, and the output iterations are performed up to the maximum number of times. Time-estimated channel

[0104] Second interference cancellation module: Based on the channel estimated by the second channel estimation module... and the signal output by the first processing unit Perform interference reconstruction and elimination, and output the signal after eliminating the previous interference.

[0105] Second symbol detection calculation module: Based on the channel estimated by the second channel estimation module... The second interference cancellation module eliminates the interference from the signal. and the signal output by the first processing unit Simultaneously, the noise variance output by the process control module is considered. With hyperparameter γ, used to detect the current block symbol The initial values ​​are set the same as those in the second channel estimation module;

[0106] Process control module: Based on the channel output of the second channel estimation module The symbols output by the second detection symbol calculation module and the signal output by the first processing unit Update noise variance Along with the hyperparameter γ, the updated value is returned to the second channel estimation module and the second symbol detection calculation module to iterate again until the iteration condition is met or the iteration number is reached. At this point, the operation of the second processing unit ends, and the detected symbol is output.

[0107] The present invention also provides a computing processing device, comprising: one or more processors; a memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, wherein when the programs are executed by the processors, they implement the steps of the single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels as described above.

[0108] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels as described above.

[0109] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus (systems), computer devices, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0110] This invention is described with reference to a flowchart of a method according to embodiments of the invention. It should be understood that each step in the flowchart and combinations thereof can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the process. Figure 1 A device for a function specified in one or more processes.

[0111] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 The function specified in one or more processes.

[0112] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 Steps of a specified function in one or more processes.

[0113] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A single-carrier underwater acoustic communication method under deep-sea long-delay clustered channels, applied at the receiving end, characterized in that, The method consists of two parts. The first part obtains the initial values ​​of the transmitted symbols through preprocessing. The second part obtains the detected symbols through joint channel estimation, interference cancellation, and symbol detection. The first part includes steps 1-6: Step 1: Receive the passband signal The baseband received signal is obtained after demodulation and downsampling. ; n represents the nth received signal. Indicates the total number of received signal blocks; Step 2: Based on the passband signal from Step 1 It performs a matched correlation with the hyperbolic FM signal used as a synchronization signal during transmission, and outputs the matched correlated signal. ; Step 3: Based on the matched correlation signal output in Step 2 The specific structure of the channel is determined by detecting the position of the synchronization peak, and the channel structure parameters are output. The specific method is as follows: Based on the matched and correlated signal output in step 2 The location of the peak is detected to determine the time interval between the arrival at the two cluster channels and the length of each cluster channel; the interval between the two cluster channels is then used to determine the location of the peak. Size of the block Based on the relationship, the number of blocks traversed by the channel is calculated. and channel length margin less than one block length , ; Step 4: Based on the baseband received signal output in Step 1 The training part and the channel structure parameters output in step 3 It determines the block used for channel estimation, the length of the corresponding training sequence, and the length of the estimated channel, and outputs the estimated channel through which the nth block symbol passes. ; Step 5: Based on the baseband received signal output in Step 1 and the channel output in step 4 Interference cancellation is performed to obtain the baseband received signal after interference cancellation. ; Step 6: Based on the channel output in Step 4 The baseband received signal after interference cancellation output in step 5 After equalization processing, the detected symbols are obtained. ; The first part concerns the received passband signal. Process steps 1 through 6 block by block, and output the first block to the last block. Block baseband received signal Channel structure parameters and the symbols after detection ; Part Two includes steps 7-10: Step 7: Based on the first block to the first block of the output from the first part Block baseband received signal Channel structure parameters and the symbols after detection Simultaneously based on detection symbols Noise variance With hyperparameters Channel estimation is performed using a Gaussian generalized approximation message-passing method based on damping factor control. In the first iteration... The initial value is the symbol obtained from the first part of the detection. The initial values ​​for noise variance and hyperparameters are set to 1, and the output iterations are performed up to the maximum number of times. Time-estimated channel ; Step 8: Based on the channel estimated in Step 7 and the baseband received signal output from the first part The system reconstructs and eliminates interference, outputting the baseband received signal after eliminating the previous interference. ; Step 9: Based on the channel estimated in Step 7 Step 8: Baseband received signal after interference elimination and the baseband received signal output from the first part At the same time, combined with noise variance With hyperparameters Detect the current block symbol The initial value is set in the same way as in step 7; Step 10: Based on the channel output in Step 7 The symbols output in step 9 and the baseband received signal output from the first part Update noise variance With hyperparameters Simultaneously, the updated value is returned to steps 7 and 9 for iteration again until the iteration stopping condition is met or the iteration count is reached. This concludes the second part, and the detected symbol is output. .

2. The method according to claim 1, characterized in that, The specific method for step 4 is as follows: Based on the baseband received signal output in step 1 The training part and the channel structure parameters output in step 3 The lengths of the training sequence and the observation data used for channel estimation are determined, and the channel length traversed by the nth symbol is estimated using a sparse Bayesian learning method. And output it.

3. The method according to claim 1, characterized in that, The specific method for step 5 is as follows: Based on the baseband received signal output in step 1 The channel output from step 4 , compared with the already detected first block symbols Jointly reconstruct inter-block interference and receive signals from baseband. After interference is eliminated, the baseband received signal after interference cancellation for the current block is obtained. .

4. The method according to claim 1, characterized in that, The maximum number of iterations in step 7 is: , No. In this iteration, the channel update process for the nth symbol is as follows: First, update the mean of the noiseless linear output variable. and variance : in , , and They are matrices , , and No. OK Column elements, , They are respectively composed of symbols , Composition , Wittoripliz matrix; , They are respectively composed of symbols , Composition , Wittoripliz matrix; and These are the posterior variance and mean of the channel to be estimated. In the first iteration, the initial values ​​are set to 1 and 0, respectively. It is an intermediate variable. At that time, the initial value is set to 0; intermediate variables The inverse residual variance is updated as follows: in This is a damping factor used to control the update rate and convergence of variables; Then the channel mean to be estimated is calculated. Variables affected by noise and its reciprocal variance : get and Then, the channel mean to be estimated is recovered from it. , , Their values ​​are as follows: At this point, the variance of the channel to be estimated is ,in yes The first in One element; When the maximum number of iterations is reached At that time, step 7 was updated.

5. The method according to claim 1, characterized in that, Step 8 specifically involves: when the channel is estimated in step 7... At that time, utilize the obtained channel With the detected symbols The interference in the nth baseband received signal is reconstructed and eliminated to obtain the nth baseband received signal after interference cancellation. .

6. The method according to claim 4, characterized in that, Specifically, step 9 involves receiving the nth baseband signal after interference elimination in step 8. And the first part of the output Block baseband received signal and the channel estimated in step 7 United Nations The baseband receives the signal, performs symbol detection, and combines this with noise variance. With hyperparameters Detect the current block symbol The initial values ​​are set in the same way as in step 7; the maximum number of iterations is set to... , No. In the next iteration The update is as follows: First update the variables. and its variance in, , and They are matrices , and The Line number Column elements, , and They are , and Composition Wittoripliz matrix; and These are the posterior variance and mean of the symbol to be detected. In the first iteration, the initial values ​​are set to 1 and 0, respectively. It is an intermediate variable. Initial value is set to 0; intermediate variable The inverse residual variance is updated as follows: Then the mean value of the symbols to be detected is calculated. Variables affected by noise and its reciprocal variance : get and Then, the mean value of the symbols to be detected is recovered from it. , , Their values ​​are as follows: At this point, the variance corresponding to the symbol to be detected is ,in It is the first of the prior variances of the symbols to be detected. 1 element; when the maximum number of iterations is reached... At that time, step 9 was updated.

7. The method according to claim 6, characterized in that, Step 10 is as follows: Based on the baseband received signal output from the first part The channel output in step 7 and the symbols output in step 9 Jointly update the The noise variance and hyperparameters of the next outer iteration, in the nth symbol block, are updated as follows: The result is then input back into steps 7 and 9 for iteration. The iteration continues until the maximum number of outer iterations is reached. If the loop termination condition is met, the iteration ends, the second part is completed, and the detected symbol is output. .

8. A single-carrier underwater acoustic communication device for deep-sea long-delay clustered channels, applied at the receiving end, characterized in that, The apparatus includes: a first processing unit for obtaining initial values ​​of transmitted symbols through preprocessing; and a second processing unit for obtaining detected symbols through joint channel estimation, interference cancellation, and symbol detection. The first processing unit includes: Signal demodulation module: for receiving passband signals The baseband received signal is obtained after demodulation and downsampling. ; n represents the nth received signal. Indicates the total number of received signal blocks; Signal matching module: based on the passband signal of the input demodulation module. It performs a matched correlation with the hyperbolic FM signal used as a synchronization signal during transmission, and outputs the matched correlated signal. ; Channel structure parameter determination module: based on the matched correlation signal output by the signal matching module. The specific structure of the channel is determined by detecting the position of the synchronization peak, and the channel structure parameters are output. The specific method is as follows: based on the matched and correlated signal output by the signal matching module. The location of the peak is detected to determine the time interval between the arrival at the two cluster channels and the length of each cluster channel; the interval between the two cluster channels is then used to determine the location of the peak. Size of the block Based on the relationship, the number of blocks traversed by the channel is calculated. and channel length margin less than one block length , ; First channel estimation module: Based on the baseband received signal output by the signal demodulation module. The training part Channel structure parameters output by the channel structure parameter determination module The algorithm determines the block used for channel estimation, the length of the corresponding training sequence, and the length of the set estimation channel, and outputs the estimated value to obtain the first... Channel through which block symbols pass ; First interference cancellation module: Based on the baseband received signal output by the signal demodulation module. and the channel output by the channel estimation module Interference cancellation is performed to obtain the baseband received signal after interference cancellation. ; First symbol detection calculation module: based on the channel output of the channel estimation module. The baseband received signal after interference cancellation output by the interference cancellation module. After equalization processing, the detected symbols are obtained. ; The first processing unit processes the received passband signal. The signal demodulation module processes each block, leading to the detection symbol calculation module, and outputs the first to the second blocks. Block receives baseband signal Channel structure parameters and the symbols after detection To the second processing department; The second processing unit includes: Second channel estimation module: Based on the first block to the second block output by the first processing unit Block receives baseband signal Channel structure parameters and the symbols after detection Simultaneously, based on the detection symbols output by the second detection symbol calculation module noise variance of the process control module output With hyperparameters Channel estimation is performed using a Gaussian generalized approximation message-passing method based on damping factor control. In the first iteration... The initial value is the symbol obtained from the first part of the detection. The initial values ​​for noise variance and hyperparameters are set to 1, and the output iterations are performed up to the maximum number of times. Time-estimated channel ; Second interference cancellation module: Based on the channel estimated by the second channel estimation module... and the baseband received signal output by the first processing unit The system reconstructs and eliminates interference, outputting the baseband received signal after eliminating the previous interference. ; Second symbol detection calculation module: Based on the channel estimated by the second channel estimation module... The second interference cancellation module eliminates the interference in the baseband received signal. and the baseband received signal output by the first processing unit Simultaneously, combined with the noise variance output by the process control module With hyperparameters Detect the current block symbol The initial values ​​are set the same as those in the second channel estimation module; Process control module: Based on the channel output of the second channel estimation module The symbols output by the second detection symbol calculation module and the baseband received signal output by the first processing unit Update noise variance With hyperparameters Simultaneously, the updated value is returned to the second channel estimation module and the second symbol detection calculation module to iterate again until the iteration stopping condition is met or the iteration count is reached. At this point, the operation of the second processing unit ends, and the detected symbol is output. .

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the single-carrier underwater acoustic communication method under the deep-sea long-delay clustered channel as described in any one of claims 1-7.

Images