Data-assisted short-wave OFDM sparse channel Turbo iterative estimation method

Abstract

The invention belongs to the field of OFDM (Orthogonal Frequency Division Multiplexing) channel estimation, and relates to a data-assisted short-wave OFDM sparse channel Turbo iterative estimation method, which comprises the following steps of: S1, receiving and processing a signal transmitted by a channel to obtain a receiving end symbol matrix Y; s2, performing channel estimation and equalization on the Y to obtain an estimated channel and an OFDM (Orthogonal Frequency Division Multiplexing) signal # imgabs0 #, performing Turbo iteration on the # imgabs1 # to obtain a symbol variance # imgabs2 #; S3, selecting a virtual pilot frequency in a signal # imgabs4 # according to the # imgabs3 #; s4, performing channel estimation and equalization on the Y according to the virtual pilot frequency to obtain a currently estimated channel and an OFDM signal # imgabs5 # S5, performing Turbo iteration on the signal # imgabs6 # to obtain a current symbol variance # imgabs7 # S6, and if a preset maximum number of iterations is reached, obtaining a final estimated channel; otherwise, returning to the step S3, and selecting the virtual pilot frequency in the signal # imgabs9 # according to the # imgabs8 #; according to the method, the virtual pilot frequency is selected from the signal # imgabs10 # according to the symbol variance, and the virtual pilot frequency is used for estimation again, so that the spectrum utilization rate is improved, and the channel estimation performance is improved.

Description

Technical Field

[0001] The present invention belongs to the field of short-wave OFDM channel estimation and relates to a data-aided short-wave OFDM sparse channel Turbo iterative estimation method. Background Art

[0002] At present, the introduction of Orthogonal Frequency Division Multiplexing (OFDM) technology in short-wave communication systems can improve the spectrum utilization efficiency and effectively reduce the impact of frequency-selective fading. The short-wave channel has the characteristic of fast time-varying, and the coherent demodulation of the short-wave OFDM system requires obtaining the channel state information through channel estimation. Therefore, channel estimation is a key issue in short-wave OFDM systems. Traditional channel estimation methods do not utilize the sparsity of short waves, resulting in low spectrum utilization efficiency and poor estimation performance. Summary of the Invention

[0003] To solve the above-mentioned problems of the prior art, the present invention adopts a data-aided short-wave OFDM sparse channel Turbo iterative estimation method, including:

[0004] S1. Receive the signal transmitted through the channel, and perform preliminary processing on the signal transmitted through the channel to obtain a received symbol matrix;

[0005] S2. Perform channel estimation and equalization on the received symbol matrix to obtain an estimated channel and OFDM signal; perform Turbo iteration on the estimated OFDM signal to obtain a symbol variance;

[0006] S3. Select virtual pilots in the estimated OFDM signal according to the symbol variance;

[0007] S4. Perform channel estimation and equalization on the received symbol matrix according to the positions of the virtual pilots to obtain the currently iteratively estimated channel and OFDM signal;

[0008] S5. Perform Turbo iteration on the currently iteratively estimated OFDM signal to obtain the current symbol variance;

[0009] S6. If the preset maximum number of iterations is reached, stop the iteration to obtain the finally estimated channel; otherwise, return to step S3 to select virtual pilots in the currently estimated OFDM signal according to the current symbol variance.

[0010] Advantageous Effects:

[0011] 1. In each iteration, the present invention selects reliable data symbols based on the CRC check result and the symbol variance of the data symbols generated by the Turbo iteration, and uses the reliable data symbols as virtual pilots. Then, the virtual pilots and the initial pilots are used for estimation again, thereby increasing the number of pilots in disguise, thereby improving the spectrum utilization and improving the channel estimation performance; 2. The present invention utilizes the sparse characteristics of the shortwave channel in the delay domain, and adopts compressed sensing to perform specific channel estimation solutions. In order to utilize the time correlation of multiple OFDM symbols through the channel, the TMSBL algorithm is used for sparse reconstruction, thereby improving the channel estimation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A schematic diagram of a data-assisted turbo iterative OFDM system provided in an embodiment of the present invention;

[0013] Figure 2 A flow chart of a data-aided shortwave OFDM sparse channel Turbo iterative estimation method provided by an embodiment of the present invention;

[0014] Figure 3 A schematic diagram of a virtual pilot provided in an embodiment of the present invention;

[0015] Figure 4 A schematic diagram of a CRC encoding structure provided in an embodiment of the present invention;

[0016] Figure 5a A schematic diagram of the effect of the number of iterations on the bit error rate in the TDACE algorithm under the HFMQ channel provided by an embodiment of the present invention;

[0017] Figure 5b A schematic diagram of the effect of the number of iterations on the bit error rate in the TDACE algorithm under the HFMM channel provided by an embodiment of the present invention;

[0018] Figure 5c A schematic diagram of the effect of the number of iterations on the bit error rate in the TDACE algorithm under the HFMD channel provided by an embodiment of the present invention;

[0019] Figure 6 A schematic diagram of BER of the TDACE algorithm under different channels provided by an embodiment of the present invention;

[0020] Figure 7a A schematic diagram of BER comparison of different algorithms under HFMQ channel provided by an embodiment of the present invention;

[0021] Figure 7b A schematic diagram of BER comparison of different algorithms under HFMM channel provided by an embodiment of the present invention;

[0022] Figure 7cSchematic diagram of BER comparison of different algorithms under the HFMD channel provided by the embodiments of the present invention;

[0023] Figure 8 Schematic diagram of comparison of running times of different algorithms provided by the embodiments of the present invention;

[0024] Figure 9 Schematic diagram of the block pilot structure used by DATAC1 provided by the embodiments of the present invention;

[0025] Figure 10 Schematic diagram of the measured comparison of the out-of-synchronization rate and the frame error rate provided by the embodiments of the present invention;

[0026] Figure 11 Schematic diagram of the measured comparison of BER provided by the embodiments of the present invention. Detailed implementation manners

[0027] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0028] Figure 1 It is a schematic diagram of a data-aided turbo iterative OFDM system, which records the entire process of the short-wave OFDM system used by this algorithm.

[0029] The present invention adopts a data-aided short-wave OFDM sparse channel Turbo iterative estimation method, which has two different operation stages, namely the initial estimation stage and the iterative estimation stage. In the initial estimation stage, the initial comb-like pilot is used for channel estimation; in the iterative estimation stage, reliable data symbols are selected as virtual pilots to perform channel estimation with the initial pilots, as Figure 2 described, and the specific steps include:

[0030] S1. Receive the signal transmitted through the channel, perform preliminary processing on the signal transmitted through the channel, and obtain the received symbol matrix Y;

[0031] Performing preliminary processing on the signal transmitted through the channel includes: removing the CP in the signal transmitted through the channel, performing serial-to-parallel conversion on the signal after removing the CP, and performing FFT on the signal after serial-to-parallel conversion to obtain the received symbol matrix Y.

[0032] S2. Perform channel estimation and equalization on the received symbol matrix Y to obtain the estimated channel and OFDM signal Perform Turbo iteration on the estimated OFDM signal to obtain the symbol variance

[0033] Channel estimation and equalization of the received symbol matrix Y include: performing TMSBL channel estimation based on the received symbol matrix to obtain the estimated channel; equalizing the received symbol matrix according to the estimated channel to obtain the estimated OFDM signal Among them, TMSBL is Temporal Multiple Sparse Bayesian Learning.

[0034] TMSBL (Temporal Multiple Sparse Bayesian Learning) is an algorithm based on sparse Bayesian learning, specifically designed to handle the Multiple Measurement Vector (MMV) problem; performing TMSBL channel estimation based on the received symbol matrix Y includes:

[0035] S21. Construct the sensing matrix Φ, the maximum number of loops r max , the termination threshold e, and the noise variance σ 2 ; Initialize the hyperparameter matrix Γ (0) = I L , the number of iterations r = 0, the positive definite time correlation matrix B (0) = I M and the hyperparameters Among them, I L , I M are identity matrices of dimensions L and M, where L is the time-domain channel length and M is the number of multi-observation vectors;

[0036] Φ is composed of the first L columns of the N×N-dimensional DFT (Discrete Fourier Transform) matrix F and the P rows corresponding to the original pilot positions in the matrix F; where L is the time-domain channel length, P is the number of original pilots, P < N, L < N, and N is the number of subcarriers; the element in the k-th row and n-th column of the N×N-dimensional DFT matrix F is

[0037] S22. Calculate the posterior covariance matrix Σ of the channel coefficients according to the current hyperparameter matrix Γ (r) , the sensing matrix Φ, and the noise variance σ 2 Calculate the posterior mean of the channel coefficients according to the received symbol matrix Y, the current posterior covariance matrix Σ, the sensing matrix Φ, and the noise variance σ 2 Calculate the posterior mean of the channel coefficients

[0038]

[0039] S23. Use the current posterior mean the posterior covariance matrix Σ, and the time correlation matrix B (r) to update the hyperparameters Get hyperparameters

[0040]

[0041] Among them, η is a positive value to ensure the positive definiteness of B, is the estimated value of B, |||| F is the F-norm.

[0042] S24, if r<r max or Then let r = r + 1 and return to step 2; otherwise terminate the iteration and get the estimated channel

[0043] S3, selecting a virtual pilot in the estimated OFDM signal according to the symbol variance;

[0044] S4: Perform channel estimation and equalization on the receiving end symbol matrix according to the position of the virtual pilot to obtain the channel and OFDM signal estimated in the current iteration

[0045] Channel estimation and equalization of the receiving end symbol matrix according to the position of the virtual pilot include: grouping the receiving end symbol matrix Y in pairs according to the time sequence to obtain multiple groups of symbol matrices y i ; The selected virtual pilots include the virtual pilots of each group of symbol matrices According to each group of symbol matrix y i and its corresponding virtual pilot The TMSBL channel estimation is performed at the position of each symbol matrix y i Estimated channel; According to the estimated channel pair corresponding to the symbol matrix y i Equalization is performed to obtain the OFDM signal estimated in the current iteration; wherein TMSBL is temporal multiple Bayesian learning.

[0046] According to the symbol matrix y i and its corresponding virtual pilot The TMSBL channel estimation at the position is similar to step S2, except that the sensing matrix Φ is changed, and the other processes are the same.

[0047] The specific steps include:

[0048] S41, according to the corresponding virtual pilot The position of the perception matrix Φ is constructed, and the maximum number of loops r is set max , termination threshold e and noise variance σ 2 ; Initialize the hyperparameter matrix Γ (0) =I L , number of iterations r = 0, time correlation matrix B (0) =IM and hyperparameters where I L and I M are identity matrices of dimensions L and M, where L is the time-domain channel length and M is the number of multi-observation vectors;

[0049] Φ consists of the first L columns of an N×N-dimensional DFT (Discrete Fourier Transform) matrix F and P rows corresponding to the positions of the original pilots and virtual pilots in matrix F 1 i.e., where FL is the matrix formed by the first L columns of matrix F, FP is the P rows corresponding to the positions of the original pilots in matrix F, is the P rows corresponding to the positions of the virtual pilots in matrix F v P 1 = P + P v , P is the number of original pilots, and P v is the number of virtual pilots.

[0050] S42. Calculate the posterior covariance matrix Σ of the channel coefficients according to the current hyperparameter matrix Γ (r) , sensing matrix Φ, and noise variance σ 2 . Calculate the posterior mean of the channel coefficients according to the current symbol matrix y i , posterior covariance matrix Σ, sensing matrix Φ, and noise variance σ 2 S43. Update the hyperparameters

[0051] using the current posterior mean , posterior covariance matrix Σ, and time correlation matrix B (r) to obtain the hyperparameters

[0052] S44. If r < r max or then set r = r + 1 and return to step S42; otherwise, terminate the iteration and obtain the symbol matrix y i the estimated channel

[0053] S5. Perform Turbo iteration on the currently iteratively estimated OFDM signal to obtain the current symbol variance where t is the number of iterations;

[0054] ​Performing Turbo iteration on the OFDM signal estimated in the current iteration includes: performing symbol demapping on the OFDM signal estimated in the current iteration to obtain the signal after symbol demapping and its log-likelihood ratio LLR(c n ); performing deinterleaving on the signal after symbol demapping and its log-likelihood ratio LLR(c n ) to obtain the deinterleaved signal and its likelihood ratio information LLR(d n ); performing decoding on the deinterleaved signal and its likelihood ratio information LLR(d n ) to obtain the decoded signal and its likelihood ratio information LLR′(d n ); performing interleaving on the decoded signal and its likelihood ratio information LLR′(d n ) to obtain the interleaved signal and its likelihood ratio information LLR′(c n ); performing symbol mapping on the interleaved signal and its likelihood ratio information LLR′(c n ) to obtain the symbol variance of the estimated OFDM signal where c n is the bit corresponding to the constellation point of subcarrier n of the signal after symbol demapping, and d n is the bit corresponding to the constellation point of subcarrier n of the deinterleaved signal.

[0055] Performing symbol demapping on the OFDM signal estimated in the current iteration includes:

[0056] Mapping the estimated OFDM symbol to its corresponding log-likelihood ratio (Log-Likelihood Ratio, LLR) for use in the maximum a posteriori decoder. Since the posterior probability of the estimated OFDM symbol is approximately P(d n,j ∣y), the log-likelihood ratio LLR(c n ) is defined as:

[0057]

[0058] where j′ = 1, 2, j = 1, 2 represent the indices of the two cases of 0 and 1 bits, d n,j represents the bit corresponding to the constellation point of subcarrier n, represents 's posterior probability, p(d n ) represents the probability that the bit corresponding to the constellation point of subcarrier n is 1 or 0. The first term of Equation (1) is defined as the extrinsic LLR, i.e., L e (d n,j ), and the second term is defined as the prior LLR, i.e., L f (d n,j ). It should be noted that Le (d n,j ) and L f (d n,j ) are strictly independent.

[0059] Similarly, the likelihood ratio information LLR′(c n ) includes the extrinsic likelihood ratio L′ e (d n,j ) and the prior likelihood ratio L′ f (d n,j ); symbol mapping for the interleaved signal and its likelihood ratio information LLR′(c n ) includes:

[0060]

[0061] where, is the OFDM signal estimated in the current iteration, sgn() is the sign function, tanh() is the hyperbolic tangent function, s k is the symbol of the k-th constellation point, L′ e (d n,j ) is the extrinsic likelihood ratio of LLR′(c n ), L′ f (d n,j ) is the prior likelihood ratio of LLR′(c n ), i represents the imaginary unit, and P() represents probability.

[0062] S6. If the maximum number of iterations T (T = 3) is reached, stop the iteration and obtain the finally estimated channel; otherwise, go back to step S3 and select virtual pilots according to the current symbol variance in the OFDM signal estimated in the current iteration.

[0063] The virtual pilot structure is as Figure 3 shown. Selecting virtual pilots in the OFDM signal estimated in the current iteration according to the current symbol variance includes:

[0064] S61. Group the OFDM symbols in the OFDM signal estimated in the current iteration into pairs in chronological order to obtain multiple groups of OFDM symbols where i is the index of the group number;

[0065] S62. Perform CRC encoding on each group of OFDM symbols to obtain two cyclic redundancy check codes for each group of OFDM symbols where j is the index of the cyclic redundancy check code;

[0066] Figure 4 ​The CRC encoding structure is shown as follows. In each group of OFDM symbols, two groups of CRC encodings are performed to obtain cyclic redundancy check codes respectively. and Each group of CRC is composed of the bits in the corresponding data symbols. When using the TMSBL algorithm, two OFDM symbols are processed each time, that is, one group of OFDM symbols. In order to perform channel estimation using virtual pilots in the MMV model, when selecting virtual pilots, all the bits corresponding to the data symbols on the same subcarrier must be selected. Therefore, several subcarriers are selected for each group of CRC for all the data symbols. To ensure the performance during compression and reconstruction, all pilot positions should be randomly distributed as much as possible. Therefore, the two groups of CRC do not select consecutive subcarriers, but select the data symbols on the crossed subcarriers in the OFDM symbols. as one group of CRC. i,1 The remaining data symbols are used as the other group of CRC. i,2 .

[0067] S63. By performing CRC checks on the two groups of cyclic redundancy check codes of each group of OFDM symbols, the virtual pilots of each group of OFDM symbols are obtained. The virtual pilots of the OFDM symbols are obtained by performing checks on the two groups of cyclic redundancy check codes

[0068]

[0069] Specifically, the virtual pilots of the OFDM symbols include: performing CRC checks on the two groups of cyclic redundancy check codes respectively. If at least one group of cyclic redundancy check codes passes the check, the data symbols corresponding to the cyclic redundancy check code that passes the check in the OFDM symbol v

[0069] v v = 15.

[0069] ​Select OFDM symbol Symbol variance in The smallest P v Data symbols on subcarriers as OFDM symbols The virtual pilot of includes: taking the OFDM symbol Adding the symbol variances of two data symbols on each subcarrier in, and taking the smallest P v Group, and taking the smallest P v The corresponding data symbols of the group serve as virtual pilots

[0070] In one embodiment, the parameters used in the simulation are shown in Table 1.

[0071] Table 1 System simulation parameter table

[0072]

[0073]

[0074] In the simulation, the short-wave channel used refers to the mid-latitude short-wave channel recommended by ITU-R F.1487, and the parameters are shown in Table 2. This channel is divided into three conditions: quiet condition (HFMQ), medium condition (HFMM), and interference condition (HFMD).

[0075] Table 2 Mid-latitude short-wave channel parameter table

[0076]

[0077]

[0078] The simulation index uses the bit error rate, which refers to the ratio of the number of decoded error bits to the total number of transmitted bits. First, study the performance changes of the method proposed by the present invention under different numbers of iterations. Figures 5(a), 5(b), and 5(c) respectively show the performance changes of channel estimation indirectly represented by the bit error rate under different numbers of iterations in the HFMQ, HFMM, and HFMD channels. It can be seen that regardless of which channel, the relative performance improvement in the second iteration is the largest. When the number of iterations reaches 3 times, the performance improvement in subsequent iterations is small. To save simulation time, the subsequent number of iterations is taken as 3 times.

[0079] Figure 6 For the performance comparison of the algorithm proposed by the present invention under the three channel conditions of HFMQ, HFMM, and HFMD. From Figure 6 It can be seen that the performance is the best under the HFMQ channel condition, followed by the HFMM channel, and the worst is the HFMD performance. This is consistent with the channel parameter results shown in Table 2. Since the delay and frequency spread of HFMQ are small, the channel condition is the best and the TDACE performance is the best.

[0080] To verify the effectiveness of the method (TDACE) of the present invention, a comparison with traditional algorithms was made. The present invention selected to compare with the Least Square (LS), Stagewise Weak Orthogonal Matching Pursuit (SWOMP), and Orthogonal Matching Pursuit (OMP) algorithms. Among them, SWOMP and OMP are greedy reconstruction algorithms in compressive sensing. The simulations of all algorithms used the same coding method and the same number of pilots. Figures 7(a), 7(b), and 7(c) are respectively the performance comparison diagrams of different algorithms under the HFMQ, HFMM, and HFMD channels. First, it can be concluded that for any algorithm, since the HFMQ channel condition is the best and the HFMD channel condition is the worst, the performance is the best under the HFMQ channel and the worst under the HFMD channel. Due to the small number of pilots, the LS algorithm cannot effectively cope with the fast time-varying characteristics of short waves, and its effect is poor. Compared with the OMP algorithm, although the performance of the SWOMP algorithm is slightly inferior, the OMP algorithm requires prior information on the channel sparsity, while the SWOMP algorithm does not require this prior information. The method proposed in the present invention is the best, and its performance is 2 dB better than that of the OMP algorithm. Since the method proposed in the present invention uses the TMSBL algorithm, it also does not require the known channel sparsity.

[0081] To compare the computational complexity of the method proposed in the present invention with other algorithms, the single-run time of each algorithm at different signal-to-noise ratios was statistically analyzed. The software used was MATLAB R2020b, and the computer CPU was the Intel 12th generation Core i7-12700 processor with a main frequency of 2.10 GHz. From Figure 8 it can be seen that as greedy reconstruction algorithms, the computational complexity of OMP and SWOMP is relatively low, only slightly higher than that of the LS algorithm. Since the SWOMP algorithm needs to solve for the sparse positions, its computational complexity is higher than that of the OMP algorithm. Since the method proposed in the present invention needs to perform iterations and solve soft information, its computational complexity is higher than that of other algorithms, and the single-run time is about 4 times that of the LS algorithm. Since the change in the signal-to-noise ratio of the method proposed in the present invention will affect the passing of CRC, resulting in different numbers of virtual pilots selected and whether the variance of soft data symbols is calculated, the running time will change with the change in the signal-to-noise ratio.

[0082] In one embodiment, real radio stations are used for communication and analysis. The communication link is selected to be sent from Luntai, Xinjiang (84°E, 41°N) and received in Finland (24°E, 68°N). The communication distance is about 5000 kilometers. The test time is from 14:00 on December 30, 2024 to 21:00 on December 30, 2024, for a total of 8 hours, with measurements taken every hour. In shortwave communication, the available frequencies vary for different paths and at different times. Therefore, frequency prediction is required. The Voice of America Coverage Analysis Program (VOACAP) software is used for frequency prediction to select a passable frequency band. The frequencies tested every hour are shown in Table 3.

[0083] Table 3 Test Frequency Table

[0084]

[0085] During the actual measurement, a comparison is made with the commonly used DATAC1 mode in the project. In this mode, the channel estimation algorithm is LS, which is different from the method proposed in the present invention that uses comb-like pilots. The DATAC1 mode uses block pilots, that is, the pilots are placed on all subcarriers of an OFDM symbol. The pilot time interval is 5, that is, there is 1 OFDM symbol with all pilots placed in every 5 OFDM symbols, and the remaining 4 OFDM symbols are used to place data symbols. See Figure 9 for details. When testing separately every hour, 10 frames of data are sent for each. Each frame in TDACE contains 136 OFDM symbols, and each frame in DATAC1 contains 135 OFDM symbols. The TDACE algorithm parameters are shown in Table 1, and the other same actual measurement parameters are shown in Table 4.

[0086] Table 4 System Actual Measurement Parameter Table

[0087]

[0088] The main actual measurement indicators include the out-of-step rate, frame error rate, and bit error rate. Among them, the out-of-step rate refers to the ratio of the number of frames that are not synchronized to the total number of frames sent, and the frame error rate refers to the ratio of the number of frames with decoding errors to the total number of frames sent. Figure 10 Fig. is the test result graph of the out-of-step rate and frame error rate. It can be seen from the figure that due to the use of the same synchronization algorithm, the out-of-step rates of the method proposed in the present invention and DATAC1 are basically the same. The frame error rate of the method proposed in the present invention is lower than that of DATAC1 at 14:00, 18:00, and 21:00. At other time points, the frame error rates of the two are the same, indicating that the performance of the method proposed in the present invention is superior to that of DATAC1. The out-of-step rate and frame error rate are relatively high at 14:00 because this is the sunrise time, and the ion activity in the ionosphere is more intense, and the channel changes rapidly.

[0089] Figure 11 This is the measured BER result. It can be seen from the figure that the test effect at 19:00 is the best, and the BER (bit error rate) of both DATAC1 and the method proposed in the present invention is 0. The method proposed in the present invention adopts three iterations, and it can be seen that the effects after three iterations are better than those of DATAC1, which proves the effectiveness of the method proposed in the present invention.

[0090] The above-mentioned embodiments further elaborate on the purpose, technical solution and advantages of the present invention. It should be understood that the above-mentioned embodiments are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made to the present invention within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A data-aided shortwave OFDM sparse channel Turbo iterative estimation method, characterized in that: include: S1, receiving the signal transmitted through the channel, performing preliminary processing on the signal transmitted through the channel, and obtaining a receiving end symbol matrix; S2, performing channel estimation and equalization on the receiving end symbol matrix to obtain an estimated channel and OFDM signal; performing Turbo iteration on the estimated OFDM signal to obtain a symbol variance; S3, selecting a virtual pilot in the estimated OFDM signal according to the symbol variance; S4, performing channel estimation and equalization on the receiving end symbol matrix according to the position of the virtual pilot to obtain the channel and OFDM signal of the current iteration estimation; S5, performing Turbo iteration on the OFDM signal estimated by the current iteration to obtain the current symbol variance; S6. If the preset maximum number of iterations is reached, the iteration is stopped to obtain the final estimated channel; Otherwise, return to step S3 to select a virtual pilot in the currently estimated OFDM signal according to the current symbol variance.

2. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 1, characterized in that: The preliminary processing of the signal after the channel transmission includes: removing the cyclic prefix CP in the signal after the channel transmission, performing serial-to-parallel conversion on the signal after the CP is removed, and performing fast Fourier transform on the signal after the serial-to-parallel conversion to obtain a receiving end symbol matrix.

3. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 1, characterized in that: Selecting a virtual pilot in the OFDM signal estimated in the current iteration includes: S61, the OFDM signal estimated in the current iteration is estimated in chronological order The OFDM symbols in the 2-bit OFDM signal are grouped in pairs to obtain multiple groups of OFDM symbols. Among them, t is the number of iterations, and i is the index of the group; S62, for each group of OFDM symbols Perform CRC encoding to obtain each group of OFDM symbols Two sets of cyclic redundancy check codes Wherein, j is the index of the cyclic redundancy check code; S63, by each group of OFDM symbols Two sets of cyclic redundancy check codes Check and get each group of OFDM symbols Virtual pilot 4. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 3, characterized in that: For each group of OFDM symbols CRC encoding includes: Select the data symbols on the interleaved subcarriers as a group of CRC i,1 , and use the remaining data symbols as another set of CRC i,2 .

5. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 3, characterized in that: Through the OFDM symbol Two sets of cyclic redundancy check codes Check and get OFDM symbol The virtual pilot includes: two sets of cyclic redundancy check codes Check them separately. If there is at least one set of cyclic redundancy check codes If the verification is passed, the OFDM symbol Cyclic redundancy check code passed in the verification The corresponding data symbol is used as OFDM symbol Virtual pilot Otherwise, select OFDM symbol The P with the smallest sign variance v OFDM symbols Virtual pilot Among them, P v is the preset number of virtual pilots.

6. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 3, characterized in that: Channel estimation and equalization of the receiving end symbol matrix according to the position of the virtual pilot include: grouping the receiving end symbol matrix in pairs according to the time sequence to obtain multiple groups of symbol matrices y i ; According to each group of symbol matrix y i and its corresponding virtual pilot The TMSBL channel estimation is performed at the position of each symbol matrix y i Estimated channel; According to the estimated channel pair corresponding to the symbol matrix y i Equalization is performed to obtain the OFDM signal estimated in the current iteration; wherein TMSBL is time series multiple sparse Bayesian learning.

7. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 6, characterized in that: According to the symbol matrix y i and its corresponding virtual pilot The location of TMSBL channel estimation includes: S41, according to the symbol matrix y i The corresponding virtual pilot The position of the perception matrix Φ is constructed, and the maximum number of loops r is set max , termination threshold e and noise variance σ 2 , initialize the hyperparameter matrix Γ (0) =I L , number of iterations r = 0, time correlation matrix B (0) =I M And hyperparameters Among them, I L ,I M is a unit matrix of dimensions L and M, where L is the time domain channel length and M is the number of multi-observation vectors; S42, according to the current hyperparameter matrix Γ (r) , perception matrix Φ and noise variance σ 2 Calculate the posterior covariance matrix Σ of the channel coefficients, based on the current symbol matrix y i , the posterior covariance matrix Σ, the perception matrix Φ and the noise variance σ 2 Calculate the posterior mean of the channel coefficients S43. Using the current posterior mean The posterior covariance matrix Σ and the time correlation matrix B (r) Update hyperparameters Get hyperparameters S44, if r<r max or Then let r=r+1 and return to step S42; otherwise, terminate the iteration and obtain the symbol matrix y i Estimated Channel 8. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 1, characterized in that: The Turbo iteration of the OFDM signal estimated by the current iteration includes: performing symbol demapping on the OFDM signal estimated by the current iteration to obtain the symbol demapped signal and its log-likelihood ratio information LLR (c n ); the signal after symbol demapping and its log-likelihood ratio information LLR (c n ) to deinterleave and obtain the deinterleaved signal and its likelihood ratio information LLR (d n );The deinterleaved signal and its likelihood ratio information LLR (d n ) is decoded to obtain the decoded signal and its likelihood ratio information LLR′(d n ); the decoded signal and its likelihood ratio information LLR′(d n ) to obtain the interleaved signal and its likelihood ratio information LLR′(c n ); for the interleaved signal and its likelihood ratio information LLR′(c n ) to perform symbol mapping and obtain the estimated symbol variance of the OFDM signal Among them, c n is the bit corresponding to the constellation point of subcarrier n of the signal after symbol demapping, d n is the bit corresponding to the constellation point of subcarrier n of the deinterleaved signal, and t is the number of iterations.

9. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 8, characterized in that: The interleaved signal and its likelihood ratio information LLR′(c n ) to perform symbol mapping including: According to the likelihood ratio information LLR′(c n ) Calculate the symbol mean of the estimated OFDM signal According to the sign mean Compute the signed variance in, is the OFDM signal estimated in the current iteration, s k is the symbol of the kth constellation point.

10. The data-aided shortwave OFDM sparse channel Turbo iterative estimation method according to claim 9, characterized in that: Likelihood ratio information LLR′(c n ) includes the external likelihood ratio L′ e (d n,j ) and the prior likelihood ratio L′ f (d n,j ) ; Calculate the symbol mean include: Where j = 1, 2 represents the index of bit 0 and bit 1, d n,j is the bit corresponding to the constellation point of subcarrier n, tanh() is the hyperbolic tangent function, i represents the imaginary unit, and P() represents the probability.

Images