CN115208483A - Underwater acoustic communication method under polar impulse interference - Google Patents

Abstract

The invention provides an underwater acoustic communication method under polar impulse interference, which modulates communication information to a carrier phase based on a single carrier phase shift keying modulation system, and carries out channel coding on information bits through a convolution encoder. The decoding process of the communication signal at the receiving end is as follows: firstly, pulse detection is carried out, pulse interference in a received signal is reconstructed, then, through a method of iterative pulse interference estimation elimination and frequency domain equalization, interference reconstruction and elimination are carried out on the received signal under the least square criterion, frequency domain equalization is carried out, and then, on the basis of equalized symbols, a channel, pulse interference and symbol joint estimation algorithm based on variational Bayes is adopted, and probability distribution of a plurality of parameters is used for fitting joint probability distribution to obtain estimated values of the channel, the interference and the symbols. The invention has the advantages that (1) the sparsity of the channel and the impulse interference is considered at the same time; (2) good robustness in an impulse noise environment; and (3) the computation complexity is acceptable while the performance is ensured.

Description

An underwater acoustic communication method under polar pulse interference

technical field

The invention relates to the field of underwater acoustic communication, in particular to a method with robust performance and acceptable computational complexity under the pulse interference under polar ice.

Background technique

Due to the rupture or extrusion of the ice layer, there is a large amount of impulse noise in the Arctic. In addition, impulse noise also appears in areas with frequent underwater biological activities. Its existence makes the background noise statistically no longer obey the Gaussian distribution, which seriously affects the traditional single-carrier frequency. Therefore, a robust single-carrier communication method is urgently needed.

The research on pulse interference suppression mainly focuses on the OFDM system. As one of the preferred solutions to realize high-speed underwater acoustic communication, single-carrier modulation has advantages compared with the OFDM system in terms of peak-to-average power ratio and anti-carrier frequency offset. There are fewer impulse interference suppression methods in the single-carrier system. For the single-carrier block transmission system, the traditional frequency-domain equalization structure is established when the background noise is Gaussian white noise, and the impact of impulse interference on it is not considered, which seriously affects the communication performance.

How to perform high-performance decoding of single-carrier communication data in underwater acoustic channels with pulse interference has become an important research topic in underwater acoustic communication technology.

SUMMARY OF THE INVENTION

The purpose of the present invention is to improve the robustness of the underwater acoustic single-carrier communication system under the polar pulse interference, and now provides a communication method for the underwater acoustic single-carrier block transmission system under the polar pulse interference. The basic idea is to design a signal processing method under the single-carrier block transmission structure, mainly through the method of iterative interference estimation cancellation combined with frequency domain equalization and time domain channel, interference, and symbol joint estimation to achieve the suppression of impulse interference and improve communication. reliability.

The object of the present invention is achieved in this way:

An underwater acoustic communication method under polar pulse interference. Based on a single-carrier phase shift keying modulation system, the communication signal is modulated to the carrier phase, and the information bits are channel-coded by a convolutional encoder. The decoding process of the communication signal at the receiving end is as follows :

Step 1: Signal preprocessing, output the signal y _p of the training part and the data signal y;

并输出；Step 2: Estimate the channel in the current symbol block based on the training part signal y _p output in step 1 and the transmitted training sequence

and output;

输出其位置选择矩阵P和对应位置处的干扰样本e；Step 3: Based on the data signal y output in step 1, perform pulse interference detection and parameterization, and parameterize the interference i as

Output its position selection matrix P and the interference sample e at the corresponding position;

Step 4: Perform pulse interference estimation and cancellation to obtain the signal z after interference cancellation;

Step 5: Perform frequency domain equalization and group phase correction based on the signal z obtained in step 4 after pulse interference cancellation, and output the time domain symbol after this equalization;

Step 6: Determine whether the maximum number of iterations is reached, if not, repeat steps 4 and 5 until the number of iterations is met, end the iteration, and output the time-domain symbol after equalization

采用基于变分贝叶斯的信道、脉冲干扰和符号联合估计算法，利用多个参数的概率分布拟合联合概率分布，得到信道、脉冲干扰和符号的估计符号D；Step 7: Based on the equalized time-domain symbols output in Step 6

Adopt the joint estimation algorithm of channel, impulse interference and symbol based on variational Bayes, and use the probability distribution of multiple parameters to fit the joint probability distribution to obtain the estimated symbol D of channel, impulse interference and symbol;

Step 8: Decode the estimated symbol D output in Step 7.

和第i-1次迭代的频域均衡输出的符号重构数据信号，随后使用最小二乘准则得到噪声样本的估计值

与步骤3输出的位置选择矩阵P相乘并从步骤1输出的数据信号y中消减，得到干扰消除后的信号z。Described step 4 and step 5 constitute impulse interference estimation and elimination and step 5 together constitute an iterative structure, in the ith iteration, based on the channel estimation result output in step 2

and the symbol reconstructed data signal of the frequency domain equalization output of the i-1th iteration, and then use the least squares criterion to obtain the estimated value of the noise sample

It is multiplied by the position selection matrix P output in step 3 and subtracted from the data signal y output in step 1 to obtain the signal z after interference cancellation.

作为已知符号进行联合估计，利用信道和干扰的稀疏性，采用变分贝叶斯将其各自的概率分布拟合为联合概率，联合估计信道、干扰、符号，在该步骤中，

表示需要被估计的超参数，分别为信道

干扰i、控制稀疏度的参数ε和噪声方差σ²，其中，ε分为ε_0:L-1、

两部分，L和K分别为信道的维度和时域符号的维度，信道

干扰i分别服从均值为0、方差为ε_0:L-1、

的高斯分布，采用变分贝叶斯将所有超参数的联合概率分布因式分解，分别更新每个超参数的分布函数

q(i)、q(ε)、q^new(σ²)，下面分别描述四个分布函数的更新：The step 7 is specifically: first, the equalized time domain symbols output in step 6 are

As a known symbol for joint estimation, the sparseness of the channel and interference is used, and variational Bayes is used to fit their respective probability distributions to a joint probability, and the channel, interference, and symbol are jointly estimated. In this step,

Represents the hyperparameters that need to be estimated, respectively the channel

Interference i, parameter ε that controls sparsity and noise variance σ ² , where ε is divided into ε _0:L-1 ,

Two parts, L and K are the dimension of the channel and the dimension of the time-domain symbol, respectively, the channel

The interference i obeys the mean value of 0, the variance is ε _0:L-1 ,

The Gaussian distribution of , uses variational Bayes to factorize the joint probability distribution of all hyperparameters, and updates the distribution function of each hyperparameter separately

q(i), q(ε), q ^new (σ ² ), the updates of the four distribution functions are described below:

(1) q(σ ² ) obeys the gamma distribution with parameters c+K-L+1 and d. Based on the data signal y output in step 1, the update result of σ ² is

Among them, x is the matrix composed of the estimated symbols. In the first joint estimation, the estimated symbols are

服从均值为μ_h，方差为∑_h的高斯分布，通过更新分布函数均值和方差，即可获得信道分布函数的更新，

的更新结果为：(2)

It obeys the Gaussian distribution with mean μ _h and variance ∑ _h . By updating the mean and variance of the distribution function, the update of the channel distribution function can be obtained,

The updated result is:

(3) q(i) obeys a Gaussian distribution with a mean of μ _i and a variance of ∑ _i . Similarly, by updating the mean and variance of the distribution function, the update of the interference distribution function can be obtained. The update result of q(i) is:

Among them, I is the identity matrix with the same dimension as ∑ _i ;

的伽马分布，ε中的每一个元素的更新结果为：(4) q(ε) obeys the parameters a+1 and

The gamma distribution of , the update result of each element in ε is:

Based on the estimated values μ _h and μ _i of the channel and interference obtained in steps (2) and (3), the estimation of the symbol x is obtained by solving the following problems:

After updating all elements in x, use its first column as an estimate of the symbol vector D and output it.

The present invention proposes an underwater acoustic communication method in an underwater acoustic channel where impulse interference exists. The inventiveness of the invention is embodied in that for a single-carrier block transmission system, firstly, iterative impulse interference estimation is performed to eliminate frequency domain equalization, and then based on the equalized symbols, Using the sparseness of the channel and interference, the distribution is further fitted, and the variational Bayesian method is used to re-estimate the channel, interference and symbol to further eliminate the impulse interference, which can realize high-performance decoding of single-carrier communication data under the impulse interference.

Compared with the prior art, the beneficial effects of the present invention are:

The present invention proposes an underwater acoustic communication method under impulse interference. For a single-carrier block transmission structure, the interference is initially eliminated through the steps of iterative interference elimination and frequency domain equalization. The joint distribution probability of interference is used for joint estimation of channel, interference and symbol, and the residual interference is eliminated again, and the final estimated symbol is output. Compared with the equalization method of traditional single carrier block transmission, the beneficial effects of the present invention are:

(1) Improve the reliability of communication

The communication reliability is highly related to the performance of the method. When there is impulse interference, the traditional frequency domain equalization method for block transmission is affected by noise, and the channel estimation result is extremely poor, resulting in a sharp decline in the equalization effect. The method proposed by the present invention first performs interference estimation and cancellation in the iterative process, equalizes the signal after interference cancellation, updates and corrects the result through an iterative structure, and further eliminates residual interference through joint estimation to obtain the symbol estimated results. Through the above steps, the influence of pulse interference is eliminated, the performance of the method is greatly improved, and the reliability of communication is effectively improved.

(2) Using the physical characteristics of the parameters, the calculation amount is small

The underwater acoustic channel and impulse interference have obvious sparsity, and the method proposed in the present invention utilizes this physical characteristic in joint estimation, which reduces the amount of computation.

Description of drawings

In order to illustrate the technical solutions in the embodiments of the present invention more clearly, the following briefly introduces the accompanying drawings to be used in the description of the embodiments. The accompanying drawings described below are only some embodiments of the present invention. For other skilled persons, other drawings can also be obtained from these drawings without creative work, wherein:

Fig. 1 is the flow chart of underwater acoustic communication method under polar pulse interference;

Fig. 2 is the frame structure of single carrier block transmission;

Fig. 3 is a time domain diagram of the collected krill noise;

Figure 4 is the bit error rate curve when the interference-to-noise ratio is 25dB under the background of impulse interference;

Figure 5 is the bit error rate curve when the interference-to-noise ratio is 30dB under the background of impulse interference.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings.

The present invention provides a method for underwater acoustic communication under the interference of polar pulses. The workflow of the present invention is shown in Figure 1, and the specific implementation is as follows:

Step 1: Input the received signal, preprocess the received signal, build a signal model of a single symbol block after preprocessing, and output the signal and data signal of the training part in the symbol block.

The preprocessing of the received signal described in step 1 needs to perform Doppler estimation and compensation on the received signal. For the single-carrier block transmission structure, the symbol block length is short, and it can be assumed that the channel time in a single symbol block is unchanged, and the data signal model can be established, and the training part signal _yp and the data signal y of the single symbol block are output.

Step 2: Based on the training part signal and the transmission training sequence output in Step 1, estimate the channel in the current symbol block and output it.

The channel estimation process described in step 2 performs channel estimation based on the training part signal y _p outputted in step 1 and the transmitted training sequence, and outputs the channel estimation result

Step 3: Based on the data signal output in Step 1, perform pulse interference detection, and output a position selection matrix and noise samples at corresponding positions.

输出其位置选择矩阵P和对应位置处的噪声样本e。The pulse interference detection and parameterization described in step 3 is to detect the existence of interference in the data signal y output in step 1, and parameterize the interference i as:

Output its position selection matrix P and the noise sample e at the corresponding position.

Step 4: together with Step 5, an iterative impulse interference estimation elimination and frequency domain equalization structure is formed. Based on the data signal output in step 1, the channel estimation result output in step 2, and the position selection matrix and impulse interference samples output in step 3, the data signal is reconstructed with the equalization symbol obtained from the previous iteration, and the value of the interference sample is estimated, which is converted from The data signal output in step 1 is subtracted and output.

重构数据信号，随后使用最小二乘准则得到噪声样本的估计值

与步骤3输出的位置选择矩阵P相乘并从步骤1输出的数据信号y中消减，得到干扰消除后的信号z。The impulse interference estimation and elimination described in step 4 and step 5 together form an iterative structure. In the i-th iteration, based on the channel estimation result output in step 2 and the symbol output by the frequency-domain equalization of the i-1-th iteration

Reconstruct the data signal and then use the least squares criterion to obtain an estimate of the noise sample

It is multiplied by the position selection matrix P output in step 3 and subtracted from the data signal y output in step 1 to obtain the signal z after interference cancellation.

Step 5: Perform frequency domain equalization and packet phase correction based on the signal obtained in step 4 after the pulse interference has been eliminated, and output the time domain symbol after this equalization.

In the frequency domain equalization and group phase correction steps described in step 5, based on the signal z after the interference cancellation output in step 4, the frequency domain equalization method is selected for equalization and converted to the time domain, and the symbols after the frequency domain equalization still remain unchanged. For the phase rotation caused by complete Doppler compensation, phase compensation is performed by applying the method of grouping phase correction, and the time domain symbol after this equalization is output.

Step 6: Determine whether the maximum number of iterations is reached, if not, repeat steps 4 and 5 until the number of iterations is met, end the iteration, and output the equalized time-domain symbols.

The judging step described in step 6 is the iteration number judgment of iterative interference estimation elimination and frequency domain equalization. If the maximum number of Iter is not reached, steps 4 and 5 are continued to be repeated until the equalized time domain symbol is output after the number of iterations is reached.

Step 7: Based on the equalized time-domain symbols output in step 6, taking into account the sparseness of channel and impulse interference, fit the joint probability distribution with their respective probability distributions, and perform a joint channel and interference symbol for the data signal output in step 1. Estimate, further eliminate interference, and output estimated symbols.

作为已知符号进行联合估计，不使用额外的训练序列，保证了系统的通信速率；其次利用信道和干扰的稀疏性，采用变分贝叶斯的思想，将其各自的概率分布拟合为联合概率，联合估计信道、干扰、符号，在该步骤中，

表示需要被估计的超参数，分别为信道

干扰i、控制稀疏度的参数ε和噪声方差σ²，其中，ε分为ε_0:L-1、

两部分，L和K分别为信道的维度和时域符号的维度，信道

干扰i分别服从均值为0、方差为ε_0:L-1、

的高斯分布，采用变分贝叶斯的思想，将所有超参数的联合概率分布因式分解，分别更新每个超参数的分布函数

q(i)、q(ε)、q^new(σ²)。下面分别描述四个分布函数的更新；The joint estimation step of channel, interference and symbol described in step 7 is the core of the present invention.

As a known symbol for joint estimation, no additional training sequence is used, which ensures the communication rate of the system; secondly, the sparseness of the channel and interference is used, and the idea of variational Bayes is adopted to fit their respective probability distributions as joint probability, jointly estimate the channel, interference, and symbols, in this step,

Represents the hyperparameters that need to be estimated, respectively the channel

Interference i, parameter ε that controls sparsity and noise variance σ ² , where ε is divided into ε _0:L-1 ,

Two parts, L and K are the dimension of the channel and the dimension of the time-domain symbol, respectively, the channel

The interference i obeys the mean value of 0, the variance is ε _0:L-1 ,

The Gaussian distribution of , adopts the idea of variational Bayes, factorizes the joint probability distribution of all hyperparameters, and updates the distribution function of each hyperparameter separately

q(i), q(ε), q ^new (σ ² ). The updates of the four distribution functions are described below;

(1) q(σ ² ) obeys the gamma distribution with parameters c+K-L+1 and d. Based on the data signal y output in step 1, the update result of σ ² is

Among them, x is the matrix composed of the estimated symbols. In the first joint estimation, the estimated symbols are

服从均值为μ_h，方差为∑_h的高斯分布，通过更新分布函数均值和方差，即可获得信道分布函数的更新，

的更新结果为：(2)

It obeys the Gaussian distribution with mean μ _h and variance ∑ _h . By updating the mean and variance of the distribution function, the update of the channel distribution function can be obtained,

The updated result is:

(3) q(i) obeys a Gaussian distribution with a mean of μ _i and a variance of ∑ _i . Similarly, by updating the mean and variance of the distribution function, the update of the interference distribution function can be obtained. The update result of q(i) is:

Among them, I is the identity matrix with the same dimension as ∑ _i .

的伽马分布，ε中的每一个元素的更新结果为：(4) q(ε) obeys the parameters a+1 and

The gamma distribution of , the update result of each element in ε is:

Based on the estimated values μ _h and μ _i of the channel and interference obtained in steps (2) and (3), the estimation of the symbol x is obtained by solving the following problems:

After updating all elements in x, use its first column as an estimate of the symbol vector D and output it.

Step 8: Decode the vector D output in Step 7, where D does not need to be judged, and can be directly decoded.

The advantages of the present invention will be further explained below in conjunction with the simulation experiment results. Obviously, the described embodiments are a part of the embodiments of the present invention, but not all of the embodiments, and are only used to illustrate and explain the present invention, but not limited to the present invention. Based on the simulation results in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

Simulation results:

Experimental conditions: In the simulation, the form of single-carrier block transmission is shown in Figure 2, the length of the block is 1024, and the 1/2 convolutional code is used for channel coding. The carrier center frequency is 12kHz, the bandwidth is 2kHz, and the passband sampling rate is 96kHz. Use QPSK mapping method. The block duration is 1024ms, followed by a zero guard interval of 50ms. In the simulation, the underwater acoustic channel with 4 taps is considered, the energy is [0, -4.43, -10.46, -10.46] dB, and the delay is [0, 5.2, 15, 21] ms, respectively.

The existence of a large number of krill in polar regions will generate a large amount of impulse noise and affect the communication performance. The krill noise in each area is similar. According to statistics, the probability density function of this impulse noise conforms to the SαS distribution. In the experiment, the impulse noise generated by krill in the waters near Singapore collected by the Acoustics Laboratory of National Singapore University was added. The time domain waveform is shown in Figure 3, and it can be seen that there are many spikes. will affect the performance of frequency domain equalization

Figures 4 and 5 show the bit error rate curves of each method with an INR of 25dB and 30dB under the background of krill noise. The closer the curve is to the x-axis, the smaller the bit error rate under the same signal-to-noise ratio, and the better the method performance. The structure means the iterative impulse interference estimation cancellation and frequency domain equalization structure mentioned in the present invention. As can be seen from the figure, the performance ranking of the four methods is: traditional frequency domain equalization method < method using iterative structure < method using iterative structure and sparse Bayesian joint estimation of channels and symbols < method proposed in the present invention. In addition to the traditional frequency domain equalization, the other three methods have carried out interference cancellation, so in the scene where interference exists, the performance is better than the traditional frequency domain equalization method. The processing method proposed by the present invention not only performs interference cancellation, but also based on the results of the iterative structure, considers the existence of residual interference in the joint estimation, makes use of the sparsity of impulse noise, and performs the residual interference analysis again when performing symbol estimation. Cancellation, so the performance is the best, which is comparable to the performance of time-frequency domain equalization without interference. Moreover, comparing the processing results in Figure 4 and Figure 5, when the interference-to-noise ratio increases, that is, when the interference energy is significantly enhanced, the performance of the processing method is not significantly affected, which proves that the proposed method is robust.

To sum up, the method proposed in the present invention is superior to the traditional equalization method in a single-carrier block transmission system.

The purpose of the present invention is to provide an underwater acoustic communication method under polar pulse interference. The invention discloses an underwater acoustic communication method under polar pulse interference, which belongs to the technical field of underwater acoustic communication. The present invention is realized by the following technical solutions: considering a single-carrier block transmission system, based on a single-carrier phase shift keying modulation system, the communication information is modulated onto the carrier phase, and the information bits are channel-coded by a convolutional encoder. The decoding process of the communication signal at the receiving end is as follows: First, pulse detection is performed to reconstruct the pulse interference in the received signal. Then through the method of iterative impulse interference estimation elimination and frequency domain equalization, the received signal is reconstructed, eliminated and equalized in frequency domain under the least squares criterion. Due to the existence of channel estimation error and noise, the influence of impulse interference cannot be completely eliminated. Then, based on the equalized symbols, a variational Bayesian-based joint estimation algorithm of channel, impulse interference and symbols is used to fit the joint probability distribution with the probability distribution of multiple parameters to obtain the estimated values of channel, interference and symbols. The advantages of the present invention are that (1) the sparsity of the channel and impulse interference is considered at the same time; (2) the robustness is good in the impulse noise environment; (3) the computational complexity is acceptable while the performance is guaranteed.

Claims (3)

1. An underwater acoustic communication method under polar impulse interference is characterized in that: based on a single-carrier phase shift keying modulation system, a communication signal is modulated to a carrier phase, information bits are subjected to channel coding through a convolutional encoder, and the decoding process of the communication signal at a receiving end is as follows:

step 1: signal preprocessing, outputting the signal y of the training part _p And a data signal y;

and 2, step: training part signal y output based on step 1 _p And transmitting a training sequence to estimate the channel in the current symbol block

And outputting;

and step 3: based on the data signal y output in the step 1, pulse interference detection and parameterization are carried out, and interference i is parameterized into

Outputting a position selection matrix P and an interference sample e at a corresponding position;

and 4, step 4: estimating and eliminating pulse interference to obtain a signal z after interference elimination;

and 5: performing frequency domain equalization and grouping phase correction based on the signal z obtained in the step 4 after the pulse interference is eliminated, and outputting a time domain symbol after the equalization;

and 6: judging whether the maximum iteration times is reached, if not, repeating the steps 4 and 5 until the iteration times are met, ending the iteration, and outputting the equalized time domain symbol

And 7: equalized time domain symbol based on step 6 output

Adopts a channel, pulse interference and symbol joint estimation algorithm based on variational Bayes, utilizes the probability distribution of a plurality of parameters to fit the joint probability distribution,obtaining an estimated symbol D of a channel, impulse interference and a symbol;

and 8: and decoding the estimated symbol D output by the step 7.

2. The method of claim 1, wherein the method comprises: the step 4 and the step 5 form an iterative structure together with the impulse interference estimation and elimination and the step 5, and in the ith iteration, the channel estimation result output by the step 2 is based on

And (3) symbol reconstruction data signals output by the frequency domain equalization of the (i-1) th iteration, and then obtaining an estimated value of a noise sample by using a least square criterion

Multiplying the position selection matrix P output by the step 3 and subtracting the position selection matrix from the data signal y output by the step 1 to obtain a signal z after interference elimination.

3. The underwater acoustic communication method under polar impulse interference according to claim 2, wherein: the step 7 specifically comprises the following steps:

firstly, the equalized time domain symbol output in step 6

Joint estimation is performed as a known symbol, the sparsity of the channel and the interference is utilized, variational Bayes is adopted to fit the respective probability distribution to joint probability, the channel, the interference and the symbol are jointly estimated, in the step,

representing the hyper-parameters, respectively channels, that need to be estimated

Interference i, sparsity controlling parameter epsilon and noise varianceσ ² Wherein ε is divided into _0:L-1 、

Two parts, L and K being the dimension of the channel and the dimension of the time domain symbol, respectively, the channel

Interference i obeys mean value of 0 and variance of epsilon _0:L-1 、

The Gaussian distribution of (2) is obtained by factorizing the joint probability distribution of all hyper-parameters by using variational Bayes, and the distribution function of each hyper-parameter is respectively updated

q(i)、q(ε)、q ^new (σ ² ) The following describes the updating of the four distribution functions, respectively:

(1)q(σ ² ) Obeying the gamma distribution with parameters c + K-L +1 and d, based on the data signal y, sigma output in step 1 ² Update the result to

Wherein x is a matrix formed by estimated symbols, and the estimated symbols are

(2)

Obey mean value of mu _h Variance is sigma _h The gaussian distribution of (2) can obtain the update of the channel distribution function by updating the mean and variance of the distribution function,

the update result of (1) is:

(3) q (i) obeys a mean value of mu _i Variance is sigma _i The mean and variance of the distribution function are updated to obtain the update of the interference distribution function, and the update result of q (i) is:

wherein I is dimension and sigma _i The same unit array;

(4) q (epsilon) obedience parameter is a +1 and

the update result of each element in epsilon is:

channel and interference estimation value mu obtained based on the steps (2) and (3) _h 、μ _i The estimate of the symbol x is obtained by solving the following problem:

after all elements in x are updated, the first column is output as an estimate of the symbol vector D.

Images