CN113242189B - Adaptive equalization soft information iteration receiving method combined with channel estimation - Google Patents

Abstract

The invention discloses an adaptive equalization soft information iterative reception method combined with channel estimation, which is suitable for a single carrier transmission system in the field of underwater acoustic communication. Time domain and frequency domain equalization, weighted and combined the preliminary equalization results, as the desired signal of the adaptive algorithm, and direct adaptive equalization of the signal, this method can significantly improve the decoding performance of the direct equalization stage; at the same time, using turbo The equalization structure exchanges soft information iteratively between the equalizer and the decoder in order to fully extract the error correction gain of the channel coding and gradually reduce the bit error rate of the system as a whole. Compared with other equalization methods, the invention can effectively eliminate the inter-symbol interference caused by the underwater acoustic channel, can better track the channel change with time, and has the advantage of lower computational complexity.

Description

An Iterative Reception Method of Adaptive Equalization Soft Information Combined with Channel Estimation

technical field

The present invention relates to the technical field of mobile communication. Specifically, it relates to an adaptive equalization soft information iterative reception method combined with channel estimation.

Background technique

Due to the serious multipath effect, fast time variation and limited bandwidth of underwater acoustic channel, underwater acoustic communication is challenging. In recent years, the research direction of underwater acoustic communication technology is mainly to improve the frequency band utilization and information transmission rate, and coherent communication technology has more advantages than incoherent technology. It can maintain a relatively reliable communication performance, and the complexity of the algorithm is also increasing, which affects the real-time communication.

In the underwater acoustic communication system, the process of sound waves from the sender to the receiver will be affected by refraction, reflection and scattering, resulting in multipath fading. Makes signal processing more difficult. In a single-carrier transmission system, the transmitted signal produces severe amplitude and phase distortion under the influence of the time-varying multipath channel, which introduces its own inter-symbol interference. In order to eliminate the interference between symbols, a more complex equalization technique needs to be adopted to eliminate the Inter Symbol Interference (ISI).

In the current communication system, there are two most commonly used equalization techniques, namely Channel Estimation based Turbo Equalizer (CE-TEQ) based on channel estimation and Direct Adaptation based Turbo equalization technique (Direct Adaptation based Turbo). Equalizer, DA-TEQ), the two methods have their own advantages and disadvantages, but the fundamental purpose is to eliminate ISI and improve the bit error performance of the system. In recent years, the turbo equalization structure has been widely used in underwater acoustic communication systems. More coding gains are gradually obtained through soft information iteration and the overall performance is improved. However, the iterative method increases the demodulation period, which affects the real-time performance.

In order to compromise between the complexity of the algorithm and the real-time performance, and at the same time to ensure the reliability of the equalization of the transmission system, the ISI in the signal transmission process is eliminated, so that the system can obtain a lower error rate. It is necessary to consider new receiver design methods, and it is urgent to propose an adaptive equalization soft information iterative reception method combined with channel estimation.

SUMMARY OF THE INVENTION

The purpose of the present invention is to solve the above-mentioned defects in the prior art, and provide an adaptive equalization soft information iterative reception method combined with channel estimation. This method is a receiver equalization method that compromises the complexity of the implementation method and real-time performance. For the single-carrier transmission system under the underwater acoustic channel, the method based on channel estimation is used first, and the signal is initially equalized in the time domain and frequency domain. Provide the desired signal for the direct equalization stage of adaptive equalization, and the iterative equalization stage uses the soft information iterative process between the equalizer and the decoder to obtain error correction gain and avoid information loss caused by hard decision; by improving the direct equalization stage results Therefore, the convergence speed of the iterative process is accelerated, and the overall performance of the single-carrier transmission system under the underwater acoustic channel is improved.

The purpose of the present invention can be achieved by adopting the following technical solutions:

An adaptive equalization soft information iterative reception method combined with channel estimation, the reception method comprises the following steps:

包括N_d个接收符号，每个训练序列

包含N_t个接收符号，[·]^T表示对向量进行转置，接收信号帧y以训练序列开头，数据块和训练序列相继排列，每个数据块前后各连接一个训练序列，分别称为该数据块的前向训练序列和后向训练序列，从接收信号帧y中提取第n个数据块r的前向训练序列t_n和后向训练序列t_n+1，采用信道估计算法进行冲激响应估计，分别得出估计信道

和

估计信道长度为L，即包含L个抽头系数；S1. After the receiver undergoes frame synchronization, Doppler estimation and compensation operations, it obtains a frame of received data, which is denoted as the received signal frame y. The received signal frame y includes N _max data blocks and N _max +1 training sequence, each data block

consists of N _d received symbols, each training sequence

It contains N _t received symbols, [ ] ^T represents the transposition of the vector, the received signal frame y starts with the training sequence, the data blocks and the training sequence are arranged in succession, and each data block is connected with a training sequence before and after, which are called the The forward training sequence and the backward training sequence of the data block, extract the forward training sequence t _n and the backward training sequence t _n+1 of the nth data block r from the received signal frame y, and use the channel estimation algorithm for impulse Response estimation, get the estimated channel separately

and

The estimated channel length is L, that is, it contains L tap coefficients;

和

采用最大比合并方法对数据块r进行频域均衡，得到发送符号序列

的第1个估计

发送符号序列x包含N_d个符号；S2. Extract the nth data block r from the received signal frame y, and use the estimated channel

and

The frequency domain equalization is performed on the data block r using the maximum ratio combining method, and the transmitted symbol sequence is obtained.

1st estimate of

The transmitted symbol sequence x contains N _d symbols;

和

分别求解时域中对应的前向滤波器w₁和后向滤波器w₂，前向滤波器w₁和后向滤波器w₂长度为L，再分别对数据块r进行时域均衡，得到前向时域均衡输出

和后向时域均衡输出

加权合并

和

后得到发送符号序列x的第2个估计

S3. According to the estimated channel

and

Solve the corresponding forward filter w ₁ and backward filter w ₂ in the time domain respectively, the length of the forward filter w ₁ and the backward filter w ₂ is L, and then perform time domain equalization on the data block r respectively, and obtain Forward Time Domain Equalization Output

and backward time domain equalization output

weighted merge

and

Then get the second estimate of the transmitted symbol sequence x

与第2个估计

进行等比例合并，硬判决后作为期望信号输入到自适应均衡器中，通过正向和反向自适应均衡得到正向自适应均衡输出

和反向自适应均衡输出

加权合并

和

后得到发送符号序列x的第3个估计

最后对

和

进行合并得到数据块r的均衡输出

S4. Send the first estimate of the symbol sequence x

with the 2nd estimate

Perform equal-proportion merging, input the desired signal to the adaptive equalizer after hard decision, and obtain the forward adaptive equalization output through forward and reverse adaptive equalization

and reverse adaptive equalization output

weighted merge

and

Then the third estimate of the transmitted symbol sequence x is obtained

last pair

and

Merge to get the balanced output of data block r

第k时刻的均衡符号

进行映射，得到均衡器的外信息

表示

携带的第j个比特，k＝0,…,N_d-1，j的取值范围取决于符号的调制方式，

经过解交织操作后作为先验信息

送到译码器中，b_m表示数据块携带的比特序列的第m个比特；S5, will equalize the output

Equilibrium symbol at time k

Mapping to get the outer information of the equalizer

express

The jth bit carried, k=0,...,N _d -1, the value range of j depends on the modulation mode of the symbol,

After the deinterleaving operation, it is used as a priori information

Sent to the decoder, b _m represents the mth bit of the bit sequence carried by the data block;

提取信道编码的纠错增益，输出后验信息L^D(b_m)，后验信息减去先验信息得到译码器的外信息

S6, the decoder according to the prior information

Extract the error correction gain of channel coding, output a posteriori information L ^D (b _m ), and subtract the prior information from the posterior information to obtain the outer information of the decoder

S7, decode and check the posterior information LD (b _m ), when the decoding is correct or the current iteration number Iter reaches the maximum iteration number Iter _max , the demodulation of the nth data block is completed, and then n= ⁿ +1, Iter=0, go back to step S1 to carry out the processing of the next data block, until the demodulation of all data blocks in the received signal frame y is completed; otherwise, make Iter=Iter+1, execute step S8, enter the iterative equalization stage;

进行交织处理，作为均衡器的先验信息

表示

携带的第j个比特，

是

映射得到的第k时刻先验符号，先验符号组成向量形式的先验输入

是自适应均衡器的反馈滤波器输入信号，先验符号

将与均衡输出

一起计算自适应算法的期望信号

S8, convert the external information of the decoder

Perform interleaving processing as a priori information for the equalizer

express

The jth bit carried,

Yes

The prior symbol at the k-th time obtained by the mapping, the prior symbol constitutes the prior input in the form of a vector

is the feedback filter input signal of the adaptive equalizer, the prior symbol

will be output with equalization

Calculate the expected signal of the adaptive algorithm together

和期望信号

更新滤波器系数，最后得到均衡输出

接着跳转到步骤S5，重新进入译码阶段。S9. The equalizer in the iterative equalization stage is composed of a feedforward filter and a feedback filter, using an adaptive algorithm, and using an equalized output

and expected signal

Update the filter coefficients, and finally get the equalized output

Then jump to step S5, and re-enter the decoding stage.

是由发送端的已知训练序列

经过信道后得到的，包含N_t个发送符号，从接收信号帧y中提取前向训练序列t_n，根据发送端的已知前向训练序列z_n，通过信道估计算法，求解前向的估计信道

接着提取后向训练序列t_n+1，根据发送端的已知后向训练序列z_n+1，求解后向的估计信道

Further, in step S1, the training sequence in the signal frame y is received

is the known training sequence from the sender

After passing through the channel, it contains N _t transmitted symbols. The forward training sequence t _n is extracted from the received signal frame y. According to the known forward training sequence z _n of the sender, the channel estimation algorithm is used to solve the forward estimated channel.

Next, the backward training sequence t _n+1 is extracted, and the backward estimated channel is solved according to the known backward training sequence z _n+1 of the sender.

和

从时域变换到频域，获得频域数据块R、频域信道响应

和

根据频域数据块和频域信道响应，采用式(1)表示的最大比合并方法进行频域均衡：Further, in step S2, the nth data block r is extracted from the received signal frame y, and r,

and

Transform from time domain to frequency domain, obtain frequency domain data block R, frequency domain channel response

and

According to the frequency domain data block and the frequency domain channel response, the maximum ratio combining method expressed by equation (1) is used to perform frequency domain equalization:

进行离散傅里叶逆变换，得到发送符号序列x的第1个估计

In the formula, J represents the number of frequency domain channel responses, and the frequency domain equalization output

Perform inverse discrete Fourier transform to obtain the first estimate of the transmitted symbol sequence x

根据公式(2)求解前向滤波器w₁：Further, in step S3, the forward estimated channel is used

The forward filter w ₁ is solved according to formula (2):

为噪声方差，I_M为M阶单位矩阵，其中M＝N₁+N₂+1，N₁和N₂分别为滤波器的因果部分和非因果部分，s则为信道卷积矩阵H₁的第N₂+L列，信道卷积矩阵H₁通过估计信道

构造而成，方法如下：In the formula,

is the noise variance, _IM is the M-order unit matrix, where M=N ₁ +N ₂ +1, N ₁ and N ₂ are the causal and non-causal parts of the filter, respectively, and s is the channel convolution matrix H ₁ Column N ₂ +L, the channel convolution matrix H ₁ estimates the channel by

constructed as follows:

分别表示估计信道

中的L个抽头系数，即

得到前向滤波器w₁后，前向时域均衡表示为：in,

represent the estimated channel, respectively

L tap coefficients in , i.e.

After obtaining the forward filter w ₁ , the forward time domain equalization is expressed as:

是第k时刻的前向均衡符号，求出数据块r全部时刻的均衡符号

组合成向量形式的前向时域均衡输出

再将前向的估计信道

换成后向的估计信道

求出后向滤波器w₂，计算数据块r全部时刻的后向均衡符号

组成向量形式的后向时域均衡输出

where r _k is the input signal at time k,

is the forward equalization symbol at the kth time, and find the equalization symbol at all times of the data block r

Forward time domain equalization output combined into vector form

Then the forward estimated channel

Switch to the backward estimated channel

Find the backward filter w ₂ , and calculate the backward equalization symbols at all times of the data block r

Backward time-domain equalization output in vector form

和后向时域均衡输出

进行加权合并，式中β为加权系数，计算得到发送符号序列x的第2个估计：Equation (4) is used for the forward time domain equalization output

and backward time domain equalization output

Perform weighted combining, where β is the weighting coefficient, and calculate the second estimate of the transmitted symbol sequence x:

Further, in step S4, direct equalization is performed. In the direct equalization stage, there is only a feedforward filter f, and the length of the filter is M. Like w ₁ and w ₂ , for forward adaptive equalization, it is divided into a training stage. And the decision stage, the balanced output formula of the two stages is:

为前馈滤波器的输入信号，

为第k时刻的均衡输出，滤波器系数的更新需要用到自适应算法，采用归一化最小均方(Normalized LeastMean Squares，NLMS)算法，系数更新公式如下：In the formula,

is the input signal of the feedforward filter,

For the balanced output at the k-th moment, the update of the filter coefficients needs to use an adaptive algorithm, using the Normalized Least Mean Squares (NLMS) algorithm, and the coefficient update formula is as follows:

为第k时刻期望信号，在训练阶段

为训练序列，在判决阶段

根据

和

进行硬判决得到，计算方法如下：In the formula, f _k is the feedforward filter at the kth time, ξ is the convergence factor, ε is a small positive constant,

is the expected signal at time k, in the training phase

is the training sequence, in the decision stage

according to

and

The hard judgment is obtained, and the calculation method is as follows:

和

分别是

和

第k时刻的均衡值，Q(·)运算表示对均衡符号进行硬判决；In the formula,

and

respectively

and

The equalization value at the kth moment, the Q(·) operation represents a hard decision on the equalization symbol;

组成向量形式的正向自适应均衡输出

保留正向自适应均衡的最后前馈滤波器

作为迭代均衡阶段的前馈滤波器的初始值；After the calculation of the output at all times of the nth data block is completed, the balanced output at time k=0,...,N _d -1

Forward adaptive equalization output in the form of a composition vector

The last feedforward filter that preserves the forward adaptive equalization

as the initial value of the feedforward filter in the iterative equalization stage;

与正向自适应均衡的区别在于，均衡器的输入和输出进行时间反转，保留反向自适应均衡的前馈滤波器

采用等比例合并

和

合并系数γ＝1/2，得到发送符号序列x的第3个估计：Inverse adaptive equalization uses the backward training sequence t _n+1 to solve the inverse adaptive equalization output

The difference from forward adaptive equalization is that the input and output of the equalizer are time-reversed, retaining the feedforward filter of reverse adaptive equalization

Consolidate in equal proportions

and

Combining coefficient γ=1/2, the third estimate of the transmitted symbol sequence x is obtained:

和

进行加权合并，得到直接均衡阶段的均衡输出

3 estimates of the symbol sequence x will be sent

and

Perform weighted merging to get the equalized output of the direct equalization stage

和

的加权系数。In the formula, α ₁ , α ₂ and α ₃ are estimated respectively

and

weighting factor.

映射为外信息，需要先采用时间平均的方法近似求解统计模型参数μ和δ²，μ是发送符号序列x的缩放因子，δ²则是x的方差，之后计算符号的概率值

a_i是发送符号集的第i个元素，发送符号集的符号数量取决于调制方式，之后求出均衡器输出的外信息

进行解交织操作，得到译码器的先验信息

Further, in step S5, the balanced output

To map as external information, it is necessary to use the time-average method to approximately solve the statistical model parameters μ and δ ² , where μ is the scaling factor of the transmitted symbol sequence x, and δ ² is the variance of x, and then calculate the probability value of the symbol

a _i is the i-th element of the transmitted symbol set, the number of symbols in the transmitted symbol set depends on the modulation method, and then the external information output by the equalizer is obtained

Perform the deinterleaving operation to obtain the prior information of the decoder

的指导下，译码器提取信道编码的纠错增益，输出后验信息L^D(b_m)，后验信息扣除先验信息将得到译码器的外信息

计算公式如下：Further, in step S6, in the decoder a priori information

Under the guidance of , the decoder extracts the error correction gain of the channel coding, outputs the a posteriori information L ^D (b _m ), and deducts the prior information from the posterior information to obtain the external information of the decoder

Calculated as follows:

Further, in step S7, the a posteriori information L ^D (b _m ) is judged and decoded, and whether the decoding result is correct according to the error detection code; when the decoding is correct, exit the decoding process of the current data block, and output the result. , let n=n+1, the number of iterations Iter=0, go back to step S1 to process the next data block until the received signal is processed; when the decoding fails and the current number of iterations is less than the maximum number of iterations Iter _max , let Iter=Iter+1, go to step S8, and execute the iterative operation.

进行交织处理，作为均衡器的先验信息

将第k时刻的先验信息映射为先验符号

组成先验输入

作为自适应均衡器的反馈滤波器输入信号。Further, in step S8, the external information of the decoder

Perform interleaving processing as a priori information for the equalizer

Map the prior information at time k to prior symbols

make up the prior input

Feedback filter input signal as adaptive equalizer.

和

进行初始化，反馈滤波器在直接均衡阶段没有保留系数

则置为零向量，训练阶段和判决阶段均衡器输出为：Further, in step S9, iterative equalization is performed, and the equalizer includes a feedforward filter f and a feedback filter b, using the

and

Initialized, the feedback filter does not retain coefficients in the direct equalization stage

Then it is set to a zero vector, and the output of the equalizer in the training phase and the decision phase is:

为第k时刻反馈滤波器的输入信号；对于自适应滤波器系数的更新，采用NLMS算法，反馈滤波器b_k更新公式为：In the formula, f _k is the feed-forward filter at the k-th time, b _k is the feedback filter at the k-th time, and r _k is the input signal of the feed-forward filter at the k-th time,

is the input signal of the feedback filter at time k; for the update of the adaptive filter coefficients, the NLMS algorithm is used, and the update formula of the feedback filter b _k is:

为训练序列t_n，在判决阶段，由均衡符号

和反馈的先验符号

合并后进行硬判决得到：During the training phase, the desired signal

is the training sequence t _n , in the decision stage, by the equalization symbol

and feedback a priori notation

A hard judgment after the merger results in:

组成向量，得到正向自适应均衡输出

保留最后更新的前馈滤波器

和反馈滤波器

The iterative equalization of the data block is completed, and the equalization symbols at the moment of k=0,...,N _d -1

Form a vector to get the forward adaptive equalization output

Keep the last updated feedforward filter

and feedback filter

和

进行初始化，均衡过程数据的输入和输出都进行时间反转，得到反向自适应均衡输出

保留最后更新的前馈滤波器

和反馈滤波器

为下次迭代均衡提供初始值；The feedforward filter fk' and feedback filter _b'k of the reverse adaptive equalization are based on the reserved _coefficients

and

Initialize, the input and output of the equalization process data are time-reversed, and the reverse adaptive equalization output is obtained.

Keep the last updated feedforward filter

and feedback filter

Provide an initial value for the next iteration of equalization;

和反向自适应均衡输出

采用等比例方式进行合并，得到本次迭代均衡的均衡输出

之后返回到步骤S5，进入译码阶段。For forward adaptive equalization output

and reverse adaptive equalization output

The equal proportion method is used to merge, and the balanced output of this iteration is obtained.

Then return to step S5 and enter the decoding stage.

Compared with the prior art, the present invention has the following advantages and effects:

1. The present invention provides the adaptive equalizer with an initial desired signal through preliminary equalization based on channel estimation. The desired signal of DA-TEQ is provided by its own equalization output, which is prone to error propagation. Compared with DA-TEQ, the adaptive equalizer combined with channel estimation can converge to the optimal solution faster under the guidance of the desired signal, reduce the error propagation effect caused by the equalizer itself providing the desired signal, and can reduce the error propagation effect in the direct equalization stage. Obtaining a more accurate equalization output is beneficial to reduce the bit error rate faster in the subsequent iterative equalization stage;

2. In the process of combining with channel estimation and equalization, the present invention only uses channel estimation and equalization in the direct equalization stage. Compared with CE-TEQ, this method can avoid the complex operations required to update the filter coefficients in the iterative stage. Combined with the adaptive equalization method, there is no need to perform operations such as matrix multiplication and inversion, and the amount of calculation is greatly reduced. The method can be applied in scenarios with high real-time requirements.

3. The present invention combines the advantages of DA-TEQ and CE-TEQ, and uses an adaptive algorithm to track the characteristics of channel changes during the equalization process, which is more suitable for fast time-varying underwater acoustic channels than CE-TEQ. At the same time, the initial expected signal provided by the channel estimation equalization can enable the adaptive equalization to obtain a lower bit error rate than the DA-TEQ method under the same number of iterations.

Description of drawings

1 is a flowchart of a method for iteratively receiving adaptive equalization soft information combined with channel estimation proposed by the present invention;

2 is a schematic diagram of a transmission frame structure of a method for iteratively receiving adaptive equalization soft information combined with channel estimation proposed by the present invention;

3 is a schematic diagram of a receiving frame structure of a method for iteratively receiving adaptive equalization soft information combined with channel estimation proposed by the present invention;

4 is a system structure diagram of a method for iteratively receiving adaptive equalization soft information combined with channel estimation proposed by the present invention;

FIG. 5 is a comparison diagram of the bit error rate of an adaptive equalization soft information iterative reception method combined with channel estimation proposed by the present invention and other equalization methods.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all 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.

Example

This embodiment specifically discloses an adaptive equalization soft information iterative reception method combined with channel estimation. In order to facilitate the understanding of the subsequent reception method, a brief description of the signal model of the communication system is given first. After the information bits d _k undergo coding and interleaving, baseband modulation, and shaping filtering, a transmission signal is generated:

Among them, x _n is the constellation symbol obtained by the baseband modulation of the coded bits, g( _t ) is the shaping filter, the symbol period is T, and Es represents the energy of the symbol. The signal s(t) is sent out after being modulated by the carrier.

At the receiving end, after the signal is synchronized, matched filtering and down-sampling, the baseband signal is obtained, and the underwater acoustic channel impulse response is modeled as a time-invariant finite impulse response filter, then the received signal can be expressed as:

In the above formula, the channel impulse response length is L, and the received symbol y _n includes not only the transmitted symbol x _n at the current moment, but also the symbol at the previous moment, indicating that the inter-symbol interference (ISI) is introduced after the signal passes through the channel. Gaussian white noise v _n .

An adaptive equalization soft information iterative reception method combined with channel estimation proposed in this embodiment is based on the above-mentioned signal model.

The specific parameters of the system model are: the information bits are encoded with a recursive systematic convolutional code of 1/2 code rate, the generator polynomial is (5,7), 12 kHz is selected as the carrier frequency of the system, the sampling rate is 96 kHz, and the transmitted symbols use QPSK Modulation mode, the symbol duration is about 166.67us, and its length is 16 sampling points.

Please refer to FIG. 1 , FIG. 2 , FIG. 3 and FIG. 4 , FIG. 1 is a flowchart of the method in this embodiment, FIG. 2 is a schematic diagram of a transmission frame structure of a signal in this embodiment, and FIG. 3 is a received frame in this embodiment Schematic diagram of the structure, FIG. 4 is a system structure diagram in this embodiment.

The meaning of each symbol is as follows:

n: data block label, indicating the nth data block currently being processed. In this embodiment, the initial value of n is 1.

N _max : the number of data blocks carried by the signal frame, N _max =4 in this embodiment.

Iter: the current number of iterations. In this embodiment, the initial value of Iter is 0.

Iter _max : the maximum number of iterations, iter _max =3 in this embodiment.

N _t : the symbol length of the training sequence, N _t =256 in this embodiment.

N _d : the symbol length of the data block, N _d =2048 in this embodiment.

N _s : the sum of the symbol lengths of the training sequence and the data block, N _s =N _t +N _d , in this embodiment, N _s =2304.

表示发送端的信号帧第n个训练序列。z _n :

Indicates the nth training sequence of the signal frame of the sender.

表示发送端的信号帧第n个数据块，本实施例中基带符号调制使用QPSK调制。x _n :

Indicates the nth data block of the signal frame of the transmitting end. In this embodiment, the baseband symbol modulation uses QPSK modulation.

表示接收端的信号帧第n个训练序列。t _n :

Indicates the nth training sequence of the signal frame at the receiver.

表示接收端的信号帧第n个数据块，在描述过程中，为了简化表达式，使用r代替r_n。r _n :

Indicates the nth data block of the signal frame at the receiving end. In the description process, in order to simplify the expression, r is used to replace r _n .

y: represents a complete received signal frame, including multiple data blocks and multiple training sequences.

信道的冲激响应估计值，本实施例中，不同的信道冲激响应估计值通过左下角的标号来区分。

The estimated value of the impulse response of the channel. In this embodiment, different estimated values of the channel impulse response are distinguished by the labels in the lower left corner.

L: the length of the channel impulse response, in this embodiment, L=121.

数据块符号的均衡输出。

Equalized output of data block symbols.

表示数据块的先验符号向量，由译码器输出的外信息进行交织映射得到。

The a priori symbol vector representing the data block is obtained by interleaving and mapping the outer information output by the decoder.

表示前馈均衡器，不同时刻通过右下角的标号k进行区分，滤波器长度为M，M＝N₁+N₂+1，其中N₂为滤波器的非因果部分长度，N₁为因果部分的长度，本实施例中N₁＝N₂＝60。f:

Represents a feed-forward equalizer. Different moments are distinguished by the label k in the lower right corner. The length of the filter is M, M=N ₁ +N ₂ +1, where N ₂ is the length of the non-causal part of the filter, and N ₁ is the causal part. The length of N ₁ =N ₂ =60 in this embodiment.

表示反馈均衡器，不同时刻通过右下角的标号k进行区分，滤波器长度也为M＝N₁+N₂+1，本实施例中N₁＝N₂＝60。b:

Represents a feedback equalizer. Different moments are distinguished by the mark k in the lower right corner. The filter length is also M=N ₁ +N ₂ +1, and in this embodiment, N ₁ =N ₂ =60.

In the iterative reception method for adaptive equalization soft information combined with channel estimation in this embodiment, the received signal frame y is composed of multiple data blocks, the data blocks are separated by a training sequence, and the received signal frame y is composed of a training sequence start and end, as shown in Figure 2. In this embodiment, all the data blocks of the signal frame need to be traversed in the processing process, and the data block r is balanced and decoded by using the forward and backward training sequences of the nth frame to solve the information carried by the received signal frame y, and the initial value of n is 1. As the processing progresses, the value of n will continue to increase until the end of signal demodulation.

The implementation process and system structure diagram of an adaptive equalization soft information iterative reception method combined with channel estimation are shown in Figure 1 and Figure 4, which specifically includes the following steps:

S1. Extract the training sequence t _n from the received signal frame y, that is, the forward training sequence of the nth data block, and the length of the training sequence is N _t . According to the known training sequence z _n of the sender, the matching pursuit algorithm is used to calculate the forward estimated channel

Then, the training sequence t _n+1 is extracted from the received signal frame y, that is, the backward training sequence of the nth data block, and the backward estimated channel is calculated according to the known training sequence z _n+1 of the sender.

和

从时域变换到频域，获得频域中的R、

和

频域均衡公式如下：S2. Extract the nth data block r from the received signal frame y, and use discrete Fourier transform to convert r, estimated channel

and

Transform from the time domain to the frequency domain to obtain R,

and

The frequency domain equalization formula is as follows:

进行离散傅里叶逆变换，得到发送符号序列x的第1个估计

In the formula, J represents the number of available channel impulse responses, and J=2 in this embodiment. on the equilibrium result

Perform inverse discrete Fourier transform to obtain the first estimate of the transmitted symbol sequence x

求解前向滤波器w₁：S3. The estimated channel solved according to step S1

Solve for the forward filter w ₁ :

为高斯白噪声的方差，本实施例中，使用信号帧前面的保护间隔估计得到，I_M为单位矩阵，M＝N₁+N₂+1，N₁为滤波器的因果部分，N₂为滤波器的非因果部分，N₁＝N₂＝60，信道冲激响应长度L＝121，s则为信道卷积矩阵H₁的第N₂+L＝181列，信道卷积矩阵H₁通过估计信道

构造而成，表示为In the formula,

is the variance of white Gaussian noise. In this embodiment, it is estimated by using the guard interval in front of the signal frame, IM is the identity matrix, _M =N ₁ +N ₂ +1, N ₁ is the causal part of the filter, and N ₂ is The non-causal part of the filter, N ₁ =N ₂ =60, the channel impulse response length L=121, s is the N ₂ +L=181th column of the channel convolution matrix H ₁ , the channel convolution matrix H ₁ passes through estimated channel

constructed, expressed as

After obtaining the forward filter w ₁ , the forward time domain equalization is expressed as:

表示第k时刻的前向均衡符号，r_k表示第k时刻滤波器的输入信号，根据前向滤波器w₁求出第n个数据块的全部时刻共2048个均衡符号，再组成向量形式的前向均衡输出

In the formula,

Represents the forward equalization symbol at the k-th time, and _rk represents the input signal of the filter at the k-th time. According to the forward filter w ₁ , a total of 2048 equalization symbols are obtained at all times of the n-th data block, and then form a vector Forward balanced output

改为后向的估计信道

根据公式(B)求出后向滤波器w₂。计算第n个数据块的每个时刻均衡符号

再组成向量形式的后向均衡输出

put the front

Change the estimated channel to the backward

The backward filter w ₂ is obtained from the formula (B). Calculate the equalization symbol at each moment of the nth data block

Reconstitute the backward equalized output in vector form

和后向时域均衡输出

进行合并，令β＝1/2，得到发送符号序列x的第2个估计：According to formula (D), the forward time domain equalization output is equalized in a proportional way.

and backward time domain equalization output

Combine, let β=1/2, and get the second estimate of the transmitted symbol sequence x:

(即k＝-N_t时)设置为零向量，第-N_t时刻的均衡符号

等于0。训练阶段和判决阶段的均衡输出如公式(E)所示：S4. The filter in the direct equalization stage is only the feedforward filter f _k . For forward adaptive equalization, the equalization process is divided into two stages, the training stage and the decision stage. At the beginning of the training phase

(i.e. when k=-N _t ) is set to zero vector, the equalized symbol at the -N _t time

equal to 0. The equalized output of the training phase and the decision phase is shown in Equation (E):

后，更新滤波器系数f_k，更新过程采用NLMS算法，则系数更新公式为：In the formula, _rk is the input signal of the feedforward filter at the kth time. Get the k-th time equalized symbol

Then, update the filter coefficient f _k , and the update process adopts the NLMS algorithm, then the coefficient update formula is:

为第k时刻期望信号。在训练阶段

为训练序列，在判决阶段

根据频域和时域的均衡输出进行硬判决得到，计算方法如下：In the formula, ξ is the convergence factor, ε is a small positive constant, in this embodiment, ε is 0.001,

is the expected signal at time k. in the training phase

is the training sequence, in the decision stage

It is obtained by hard decision based on the equalized output in the frequency domain and time domain. The calculation method is as follows:

和

分别是发送符号序列x的第1个估计

和第2个估计

中第k时刻的均衡符号。对于DA-TEQ而言，在判决阶段的期望信号是由自身的均衡符号

进行硬判决得到，其可信程度受限于输出符号的准确度，由于直接均衡阶段的滤波器没有可用的先验信息，该阶段的均衡输出会有较大的误差，通过这个均衡符号获得期望信号，其可信度也大大降低，容易引起后续均衡器性能的恶化，造成误差传播效应。因此，引入时域和频域均衡结果作为期望信号，由于该均衡结果的可信程度远高于自适应滤波器本身的均衡输出，能加快自适应算法的收敛过程，在直接均衡阶段获得较低的起始误码率。In the formula,

and

are the first estimates of the transmitted symbol sequence x, respectively

and the second estimate

The equalized symbol at the kth instant in . For DA-TEQ, the desired signal in the decision stage is composed of its own equalized symbols

It is obtained by hard decision, and its credibility is limited by the accuracy of the output symbol. Since the filter in the direct equalization stage has no available prior information, the equalization output of this stage will have a large error, and the expectation is obtained through this equalization symbol. The reliability of the signal is also greatly reduced, which is easy to cause the deterioration of the performance of the subsequent equalizer, resulting in the effect of error propagation. Therefore, the time-domain and frequency-domain equalization results are introduced as the desired signal. Since the credibility of the equalization results is much higher than the equalization output of the adaptive filter itself, it can speed up the convergence process of the adaptive algorithm and obtain lower results in the direct equalization stage. the starting bit error rate.

组成向量形式的正向自适应均衡输出

保留直接均衡阶段的最后的前馈滤波器

作为迭代均衡的训练阶段前馈滤波器的初始值。After the equalization is completed, the equalization symbols at the time k=0,...,N _d -1 in the decision stage

Forward adaptive equalization output in the form of a composition vector

The last feedforward filter that retains the direct equalization stage

as the initial value of the feedforward filter in the training phase of iterative equalization.

将判决阶段k＝0,…,N_d-1时刻的均衡符号

组成向量形式，再进行时间反转就得到反向自适应均衡输出

反向自适应均衡最后的前馈滤波器

也需要保留下来，迭代均衡阶段将用于初始化。Inverse adaptive equalization uses a feedforward filter f _k ′, the input signal of the filter needs to be time-reversed, and the equalized symbol at each moment is obtained through equalization

Equilibrium symbols at the time of decision stage k=0,...,N _d -1

It is composed of a vector form, and then time reversal is performed to obtain an inverse adaptive equalization output.

Inverse adaptive equalization final feedforward filter

It also needs to be preserved, the iterative equalization phase will be used for initialization.

和

令γ＝1/2，自适应均衡后的发送符号序列x的第3个估计为：Consolidate in equal proportions

and

Let γ=1/2, the third estimate of the transmitted symbol sequence x after adaptive equalization is:

和

进行合并，公式如下：Finally 3 estimates of the symbol sequence will be sent

and

To merge, the formula is as follows:

In the formula, α ₁ =α ₂ =α ₃ =1/3, and the combined result is used as the equalization output of the direct equalization stage.

分别表示发送符号的缩放因子和方差。本实施例中采用时间平均的方法计算模型参数，计算公式如下：S5. In this embodiment, it is assumed that the received symbols obey the Gaussian distribution, and the parameters of the probability model are obtained by estimation, wherein the key parameters μ _k and

represent the scaling factor and variance of the transmitted symbols, respectively. In the present embodiment, the time-averaged method is used to calculate the model parameters, and the calculation formula is as follows:

再根据贝叶斯定理计算均衡器外信息

解映射计算公式如下：After getting the model parameters, calculate the conditional probability

Then calculate the information outside the equalizer according to Bayes' theorem

The demapping calculation formula is as follows:

进行解交织，并将结果作为译码器的先验信息

external information

Perform deinterleaving and use the result as a priori information for the decoder

计算公式如下：S6. In this embodiment, the decoder adopts the BCJR algorithm based on the maximum a posteriori probability criterion to calculate the posterior information LD (b _m ), and the posterior information ^deducts the prior information to obtain the external information of the decoder

Calculated as follows:

where b _m represents the mth bit before the information bits are not interleaved.

S7. The a posteriori information ^LD (b _m ) is the likelihood probability of the information bits, and the decision decoding is performed according to its probability value. In this embodiment, the error detection code adopts the cyclic redundancy check code, which is used for the check decoding result.

When the decoding is correct, the demodulation of the nth data block is completed in advance; or when the decoding result is wrong and the maximum number of iterations _Itermax is reached, the demodulation of the nth data block is stopped. Then, let n=n+1, the number of iterations Iter=0, go back to step S1 to process the next data block, until the received signal frame processing is completed.

When the decoding result is wrong, and the current iteration number Iter is less than the maximum iteration number Iter _max , set Iter=Iter+1, enter the iteration stage of step S8, and still process the current data block.

进行交织处理得到均衡器的先验信息

再将交织后的先验信息映射为先验符号

本实施例采用QPSK调制，映射方式如下：S8. Before iterative equalization, the external information needs to be mapped into symbols and fed back to the equalizer.

Perform interleaving to obtain prior information of the equalizer

Then map the interleaved prior information into prior symbols

This embodiment adopts QPSK modulation, and the mapping method is as follows:

和

分别是先验符号

的两个比特对应的先验信息，当前数据块的所有先验符号映射完成后，组合成先验符号向量

作为迭代均衡阶段均衡器的反馈滤波器输入信号。In the formula,

and

a priori notation

The a priori information corresponding to the two bits of

Feedback filter input signal as equalizer in iterative equalization stage.

由直接均衡阶段或上次迭代均衡阶段的保留系数

决定，反馈滤波器的初始值

则置为零向量或保留的决定

取决于迭代次数Iter。训练阶段和判决阶段均衡器输出为：S9. The filter in the iterative equalization stage is composed of a feedforward filter _{fk and a feedback filter bk .} The initial value of the feedforward filter is

Retention coefficients from the direct equalization stage or the last iterative equalization stage

determine, the initial value of the feedback filter

then set to zero vector or keep the decision

Depends on the number of iterations Iter. The training phase and decision phase equalizer outputs are:

为反馈输入信号。In the formula, f _k is the feedforward filter at the kth moment, _bk is the feedback filter at the kth moment, r _k is the feedforward input signal,

Input signal for feedback.

后，本实施例采用NLMS算法对滤波器系数进行更新，前馈滤波器f_k更新公式与步骤S4的公式(F)一样，反馈滤波器b_k更新公式为：The filter equalizes the output at time k

Then, the present embodiment adopts the NLMS algorithm to update the filter _{coefficients. The update formula of the feedforward filter fk is the same as the formula (F) of step S4, and the update formula of the feedback filter bk is} :

在训练阶段为前向训练序列t_n，在判决阶段则由均衡符号

和反馈的先验符号

合并后进行硬判决得到：where the desired signal

In the training phase, it is the forward training sequence t _n , and in the decision phase, the equalization symbol

and feedback a priori notation

A hard judgment after the merger results in:

组成向量，得到正向自适应均衡输出

再保留最后更新的前馈滤波器

和反馈滤波器

Equilibrium symbols at time k=0,...,N _d -1

Form a vector to get the forward adaptive equalization output

Retain the last updated feedforward filter

and feedback filter

再将k＝0,…,N_d-1时刻的均衡符号组成向量，进行时间反转，得到反向自适应均衡输出

保留最后更新的前馈滤波器

和反馈滤波器

完成反向均衡。The coefficients of the inverse equalization filter are f _k ′ and b′ _{k respectively} . After the filter input is time-reversed, the equalization symbol at each moment is calculated as

Then, the equalization symbols at k=0,...,N _d -1 time are formed into a vector, and the time is reversed to obtain the reverse adaptive equalization output.

Keep the last updated feedforward filter

and feedback filter

Complete reverse equalization.

和

得到本次迭代均衡的均衡输出

之后返回到步骤S5，重新进入译码阶段。Finally, merge in equal proportions

and

Get the equalized output of this iteration equalization

Then return to step S5, and re-enter the decoding stage.

The implementation mode and specific parameters of the present invention are described in detail above. Next, a performance comparison will be made with two commonly used equalization methods, DA-TEQ and CE-TEQ. The channel estimation equalization algorithm adopts the linearity based on the minimum mean square error criterion. Equalization, the adaptive equalization algorithm uses the normalized least mean square algorithm. According to the previous system parameter description, N _max = 4, that is, each signal frame carries 4 data blocks, each data block contains 2048 QPSK symbols, and the code rate is 1/2, then the information bits carried in one signal frame are 8192 8,000 bits are selected to carry valid information, the remaining 192 bits are used for other purposes or reserved, and 100 frames of data signals are sent to ensure that the amount of data is sufficient. , and set the signal-to-noise ratio to 6dB, set various algorithms to go through 8 iterations in the decoding process, and finally adopt different methods to equalize and calculate the bit error rate. The result is shown in Figure 5.

As can be seen from Figure 5, the defect of DA-TEQ is that the bit error rate is high in the initial stage, but it can continuously reduce the bit error rate in the iterative equalization process, and even surpass the performance of CE-TEQ in the later stage; The advantage is that a lower bit error rate can be obtained in the initial stage, but the gain of the iterative process cannot be well extracted. After the first few iterations, the bit error platform is reached, and the bit error rate cannot be further reduced. The adaptive equalization algorithm can obtain the advantages of both, and can have a lower error starting point like CE-TEQ in the early iteration, and quickly reduce the error rate in the iterative process. In conclusion, an adaptive equalization soft information iterative reception method combined with channel estimation proposed in this embodiment has better decoding performance.

The above-mentioned embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above-mentioned embodiments, and any other changes, modifications, substitutions, combinations, The simplification should be equivalent replacement manners, which are all included in the protection scope of the present invention.

Claims (10)

1. A method for iteratively receiving adaptive equalization soft information combined with channel estimation, characterized in that the receiving method comprises the following steps: S1. After the receiver undergoes frame synchronization, Doppler estimation and compensation operations, it obtains a frame of received data, which is denoted as the received signal frame y. The received signal frame y includes N _max data blocks and N _max +1 training sequence, each data block

consists of N _d received symbols, each training sequence

It contains N _t received symbols, [ ] ^T represents the transposition of the vector, the received signal frame y starts with the training sequence, the data blocks and the training sequence are arranged in succession, and each data block is connected with a training sequence before and after, which are called the The forward training sequence and the backward training sequence of the data block, extract the forward training sequence t _n and the backward training sequence t _n+1 of the nth data block r from the received signal frame y, and use the channel estimation algorithm for impulse Response estimation, get the estimated channel separately

and

The estimated channel length is L, that is, it contains L tap coefficients; S2. Extract the nth data block r from the received signal frame y, and use the estimated channel

and

The frequency domain equalization is performed on the data block r using the maximum ratio combining method, and the transmitted symbol sequence is obtained.

1st estimate of

The transmitted symbol sequence x contains N _d symbols; S3. According to the estimated channel

and

Solve the corresponding forward filter w ₁ and backward filter w ₂ in the time domain respectively, the length of the forward filter w ₁ and the backward filter w ₂ is L, and then perform time domain equalization on the data block r respectively, and obtain Forward Time Domain Equalization Output

and backward time domain equalization output

weighted merge

and

Then get the second estimate of the transmitted symbol sequence x

S4. Send the first estimate of the symbol sequence x

with the 2nd estimate

Perform equal-proportion merging, input the desired signal into the adaptive equalizer after hard decision, and obtain the forward adaptive equalization output through forward and reverse adaptive equalization

and reverse adaptive equalization output

weighted merge

and

Then the third estimate of the transmitted symbol sequence x is obtained

last pair

and

Merge to get the balanced output of data block r

S5, will equalize the output

Equilibrium symbol at time k

Mapping to get the outer information of the equalizer

express

The jth bit carried, k=0,...,N _d -1, the value range of j depends on the modulation mode of the symbol,

After the deinterleaving operation, it is used as a priori information

Sent to the decoder, b _m represents the mth bit of the bit sequence carried by the data block; S6, the decoder according to the prior information

Extract the error correction gain of channel coding, output a posteriori information L ^D (b _m ), and subtract the prior information from the posterior information to obtain the outer information of the decoder

S7, decode and check the posterior information LD (b _m ), when the decoding is correct or the current iteration number Iter reaches the maximum iteration number Iter _max , the demodulation of the nth data block is completed, and then n= ⁿ +1, Iter=0, go back to step S1 to carry out the processing of the next data block, until the demodulation of all data blocks in the received signal frame y is completed; otherwise, make Iter=Iter+1, execute step S8, enter the iterative equalization stage; S8, convert the external information of the decoder

Perform interleaving processing as a priori information for the equalizer

express

The jth bit carried,

Yes

The prior symbol at the k-th time obtained by the mapping, the prior symbol constitutes the prior input in the form of a vector

is the feedback filter input signal of the adaptive equalizer, the prior symbol

will be output with equalization

Calculate the expected signal of the adaptive algorithm together

S9. The equalizer in the iterative equalization stage is composed of a feedforward filter and a feedback filter, using an adaptive algorithm, and using an equalized output

and expected signal

Update the filter coefficients, and finally get the equalized output

Then jump to step S5, and re-enter the decoding stage. 2. The method for iteratively receiving adaptive equalization soft information combined with channel estimation according to claim 1, wherein in step S1, the training sequence in the signal frame y is received

is the known training sequence from the sender

After passing through the channel, it contains N _t transmitted symbols. The forward training sequence t _n is extracted from the received signal frame y. According to the known forward training sequence z _n of the sender, the channel estimation algorithm is used to solve the forward estimated channel.

Then, the backward training sequence t _n+1 is extracted, and the backward estimated channel is solved according to the known backward training sequence z _n+1 of the sender.

3. a kind of adaptive equalization soft information iterative reception method combined with channel estimation according to claim 1, is characterized in that, in step S2, extracts the nth data block r from the received signal frame y, utilizes discrete Fu Lie transform transforms r,

and

Transform from time domain to frequency domain, obtain frequency domain data block R, frequency domain channel response

and

According to the frequency domain data block and the frequency domain channel response, the maximum ratio combining method expressed by equation (1) is used to perform frequency domain equalization:

In the formula, J represents the number of frequency domain channel responses, and the frequency domain equalization output

Perform inverse discrete Fourier transform to obtain the first estimate of the transmitted symbol sequence x

4. The method for iterative reception of adaptive equalization soft information combined with channel estimation according to claim 1, characterized in that, in step S3, a forward estimated channel is used

The forward filter w ₁ is solved according to formula (2):

In the formula,

is the noise variance, _IM is the M-order unit matrix, where M=N ₁ +N ₂ +1, N ₁ and N ₂ are the causal and non-causal parts of the filter, respectively, and s is the channel convolution matrix H ₁ Column N ₂ +L, the channel convolution matrix H ₁ estimates the channel by

constructed as follows:

in,

represent the estimated channel, respectively

L tap coefficients in , i.e.

After obtaining the forward filter w ₁ , the forward time domain equalization is expressed as:

where r _k is the input signal at time k,

is the forward equalization symbol at the kth time, and find the equalization symbol at all times of the data block r

Forward time domain equalization output combined into vector form

Then the forward estimated channel

Switch to the backward estimated channel

Find the backward filter w ₂ , and calculate the backward equalization symbols at all times of the data block r

Backward time-domain equalization output in vector form

Equation (4) is used for the forward time domain equalization output

and backward time domain equalization output

Perform weighted combining, where β is the weighting coefficient, and calculate the second estimate of the transmitted symbol sequence x:

5. a kind of adaptive equalization soft information iterative reception method combined with channel estimation according to claim 1, is characterized in that, in step S4, carry out direct equalization, the direct equalization stage has only feedforward filter f, and the filter's The length is M. Like w ₁ and w ₂ , for forward adaptive equalization, it is divided into a training phase and a decision phase. The equalization output formula of the two phases is:

In the formula,

is the input signal of the feedforward filter,

For the balanced output at the kth time, the update of the filter coefficients needs to use an adaptive algorithm, using the normalized least mean square algorithm, and the coefficient update formula is as follows:

In the formula, f _k is the feedforward filter at the kth time, ξ is the convergence factor, ε is a small positive constant,

is the expected signal at time k, in the training phase

is the training sequence, in the decision stage

according to

and

The hard judgment is obtained, and the calculation method is as follows:

In the formula,

and

respectively

and

The equalization value at the kth moment, the Q(·) operation represents a hard decision on the equalization symbol; After the calculation of the output at all times of the nth data block is completed, the balanced output at time k=0,...,N _d -1

Forward adaptive equalization output in the form of a composition vector

The last feedforward filter that preserves the forward adaptive equalization

as the initial value of the feedforward filter in the iterative equalization stage; Inverse adaptive equalization uses the backward training sequence t _n+1 to solve the inverse adaptive equalization output

The difference from forward adaptive equalization is that the input and output of the equalizer are time-reversed, retaining the feedforward filter of reverse adaptive equalization

Consolidate in equal proportions

and

Combining coefficient γ=1/2, the third estimate of the transmitted symbol sequence x is obtained:

3 estimates of the symbol sequence x will be sent

and

Perform weighted merging to get the equalized output of the direct equalization stage

In the formula, α ₁ , α ₂ and α ₃ are estimated respectively

and

weighting factor. 6. The method for iteratively receiving adaptive equalization soft information combined with channel estimation according to claim 1, wherein in step S5, the equalization output is

To map as external information, it is necessary to use the time-average method to approximately solve the statistical model parameters μ and δ ² , where μ is the scaling factor of the transmitted symbol sequence x, and δ ² is the variance of x, and then calculate the probability value of the symbol

a _i is the i-th element of the transmitted symbol set, the number of symbols in the transmitted symbol set depends on the modulation method, and then the external information output by the equalizer is obtained

Perform the deinterleaving operation to obtain the prior information of the decoder

7. The method for iteratively receiving adaptive equalization soft information combined with channel estimation according to claim 1, characterized in that, in step S6, in the decoder a priori information

Under the guidance of , the decoder extracts the error correction gain of the channel coding, outputs the a posteriori information L ^D (b _m ), and deducts the prior information from the posterior information to obtain the external information of the decoder

Calculated as follows:

8. a kind of adaptive equalization soft information iterative reception method combined with channel estimation according to claim 1, is characterized in that, in step S7, carries out decision decoding to a posteriori information L ^D (b _m ), and according to The error detection code judges whether the decoding result is correct; when the decoding is correct, exit the decoding process of the current data block, output the result, let n=n+1, the number of iterations Iter=0, go back to step S1 to process the next data block , until the received signal is processed; when the decoding fails and the current number of iterations is less than the maximum number of iterations Iter _max , set Iter=Iter+1, go to step S8, and execute the iterative operation. 9. The method for iterative reception of adaptive equalization soft information combined with channel estimation according to claim 1, wherein in step S8, the external information of the decoder is

Perform interleaving processing as a priori information for the equalizer

Map the prior information at time k to prior symbols

make up the prior input

Feedback filter input signal as adaptive equalizer. 10. The method for iteratively receiving soft information with adaptive equalization combined with channel estimation according to claim 1, wherein in step S9, iterative equalization is performed, and the equalizer comprises a feedforward filter f and a feedback filter b , using the last equilibrium reserved

and

Initialized, the feedback filter does not retain coefficients in the direct equalization stage

Then it is set to a zero vector, and the output of the equalizer in the training phase and the decision phase is:

In the formula, f _k is the feed-forward filter at the k-th time, b _k is the feedback filter at the k-th time, and r _k is the input signal of the feed-forward filter at the k-th time,

is the input signal of the feedback filter at time k; for the update of the adaptive filter coefficients, the normalized least mean square algorithm is used, and the update formula of the feedback filter b _k is:

During the training phase, the desired signal

is the training sequence t _n , in the decision stage, by the equalization symbol

and feedback a priori notation

A hard judgment after the merger results in:

The iterative equalization of the data block is completed, and the equalization symbols at the moment of k=0,...,N _d -1

Form a vector to get the forward adaptive equalization output

Keep the last updated feedforward filter

and feedback filter

The feedforward filter _f'k and the feedback filter _b'k of the reverse adaptive equalization are based on the reserved coefficients

and

Initialize, the input and output of the equalization process data are time-reversed, and the reverse adaptive equalization output is obtained.

Keep the last updated feedforward filter

and feedback filter

Provide an initial value for the next iteration of equalization; For forward adaptive equalization output

and reverse adaptive equalization output

The equal proportion method is used to merge, and the balanced output of this iteration is obtained.

Then return to step S5 and enter the decoding stage.

Images