CN115865101A - Variable sliding window decoding method and system of double-space coupling LDPC code - Google Patents

Abstract

The invention relates to a variable sliding window decoding method and a system of a double-space coupling LDPC code, which relate to the field of communication and comprise the following steps: initializing the decoding window width of a decoder of the double-space coupling LDPC code; BP decoding is carried out on the received code words in a current decoding window of a current decoding position, and the log-likelihood ratio of a target symbol in the current decoding window is obtained; calculating the threshold value of the log-likelihood ratio and the average value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position; judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or the width of the current decoding window reaches the maximum value of a preset window; if not, adding 1 to the width of the current decoding window, and obtaining the log-likelihood ratio of the target symbol in the current decoding window again; and if so, sliding the current decoding window to the next decoding position until all the code words received by the decoder are decoded. The invention improves the decoding performance.

Description

Variable sliding window decoding method and system of double-space coupling LDPC code

Technical Field

The invention relates to the technical field of communication, in particular to a variable sliding window decoding method and system of a double-space coupling LDPC code.

Background

In conventional communication systems, source and channel design optimization is two separate processes. The source coding is to compress the source sequence, reduce redundancy and improve the transmission effectiveness of the communication system. The channel coding reasonably adds some redundant information bit sequences to the information sequence, and improves the reliability of the communication system transmission. However, there is an independent process between the two, so the redundant information existing after source coding cannot be used in the subsequent decoding process. A decoder in a Joint Source-Channel Coding (JSCC) system can improve decoding performance by using redundant information of a Source. The decoding delay is an important index for measuring the quality of the decoder.

A spatial coupled low-density parity-check (SC-LDPC) code is introduced into a JSC system and named as a double spatial coupled LDPC code (DSC-LDPC) system. FIG. 3 is a diagram of a DSC-LDPC code, which is expanded M times to obtain a Tanner graph, where M is called an expansion factor.

The SC-LDPC code is a convolutional LDPC code. When the code length of the SC-LDPC code is long enough, a Belief-Propagation (BP) decoding threshold value can reach a maximum a posteriori decoding threshold value. For a spatially coupled LDPC code decoder, a longer coupling length will result in a larger decoding delay. The SC-LDPC code is obtained by coupling L regular LDCP codes, wherein L is defined as the coupling length, so that the check matrix of the SC-LDPC code presents a special structure of a non-zero diagonal band. Fig. 2 shows a coupling process of SC-LDPC codes, where (a) in fig. 2 is L regular LDCP codes, (b) in fig. 2 is an edge extension scheme, and (c) in fig. 2 is a coupled (3, 6, L) SC-LDPC code obtained by coupling. It is decoded using a Sliding Window Decoding (SWD) algorithm, thereby reducing decoding delay. The sliding window decoding algorithm limits BP decoding in a W-dimensional window, and only needs to receive code words in the window with the size of W to start BP decoding on the code words in the window. Fig. 4 shows that the decoding window of the SWD slides along the original pattern of the DSC-LDPC code. Although SWD reduces its decoding delay and decoding complexity, it still leaves the penalty of decoding performance to be reduced at the expense of decoding performance.

Disclosure of Invention

The invention aims to provide a variable sliding window decoding method and system of a double-space coupling LDPC code, which improve the decoding performance.

In order to achieve the purpose, the invention provides the following scheme:

a variable sliding window decoding method of double space coupling LDPC codes comprises the following steps:

initializing the decoding window width of a decoder of the double-space coupling LDPC code;

BP decoding is carried out on the received code words in a current decoding window of a current decoding position, and the log-likelihood ratio of a target symbol in the current decoding window is obtained and recorded; the current decoding position is the position of the upper left corner of the current decoding window; the target symbol is the leftmost source original model graph unit of the current decoding window;

calculating a log-likelihood ratio threshold and an average value of the log-likelihood ratios of target symbols of all decoding windows recorded at the current decoding position;

judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or the width of the current decoding window reaches the maximum value of a preset window;

if not, adding 1 to the width of the current decoding window, and returning to the step of carrying out BP decoding on the received code word in the current decoding window at the current decoding position to obtain the log-likelihood ratio of the target symbol in the current decoding window;

if so, completing decoding of the code word in the current decoding window, sliding the current decoding window to the next decoding position, returning to the step of performing BP decoding on the received code word in the current decoding window at the current decoding position to obtain the log-likelihood ratio of the target symbol in the current decoding window until all the code words received by the decoder are completely decoded.

Optionally, before initializing a decoding window width of a decoder of the double spatial coupling LDPC code, the method specifically includes:

the information source generates a random sequence obeying Bernoulli distribution, and the probability that a bit in the random sequence is 1 is 0.02;

carrying out compression coding on the random sequence to obtain a sequence after source coding;

coding the sequence after the information source coding based on the information bit and the check bit on the channel to obtain the sequence after the channel coding;

adopting BPSK modulation to modulate the sequence after the channel coding to obtain a modulated signal;

the modulated signal is transmitted to the decoder over an AWGN channel.

Optionally, the coupling length of the source is 16, the coupling width is 2, and the spreading factor is 160.

Optionally, the coupling length of the channel is 20, the coupling width is 2, and the spreading factor is 160.

Optionally, the log likelihood ratio threshold is expressed as:

where θ represents a log likelihood ratio threshold, L ^sc Representing the coupling length of the source, m _s Indicating the coupling width, W, of the source _max Represents the maximum value of the preset window, W _inc Represents window increment, W _inc ＝W _max -W _f ，W _f Which represents the initial width of the decoding window,

represents the average log-likelihood ratio of the target symbol and t represents the position of the target symbol.

Optionally, the preset window maximum value is 16, and the initial width of the decoding window is 3.

Optionally, the sliding the current decoding window to the next decoding position specifically includes:

and sliding the current decoding window to the right by one original model graph unit.

The invention discloses a variable sliding window decoding system of double space coupling LDPC code, comprising:

a decoding window width initialization module for initializing the decoding window width of the decoder of the double space coupling LDPC code;

a log-likelihood ratio determining module of the target symbol in the current decoding window, which is used for carrying out BP decoding on the received code word in the current decoding window at the current decoding position, obtaining the log-likelihood ratio of the target symbol in the current decoding window and recording the log-likelihood ratio; the current decoding position is the position of the upper left corner of the current decoding window; the target symbol is the leftmost source original model graph unit of the current decoding window;

the average value calculation module of the log-likelihood ratio is used for calculating the threshold value of the log-likelihood ratio and the average value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position;

the judging module is used for judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or whether the width of the current decoding window reaches the maximum value of a preset window;

a decoding window widening module, configured to add 1 to the width of the current decoding window and return to the log-likelihood ratio determining module of the target symbol in the current decoding window when the determining module determines that the target symbol is not in the current decoding window;

and the current decoding window sliding module is used for sliding the current decoding window to the next decoding position when the judgment module judges that the code word in the current decoding window completes decoding, and returning to the log-likelihood ratio determining module of the target symbol in the current decoding window until all the code words received by the decoder complete decoding.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects:

the invention calculates the log-likelihood ratio threshold and the average value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position, and the size of the decoding window is changed according to the average value of the log-likelihood ratios of the target symbols of the information source, thereby improving the decoding performance.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.

FIG. 1 is a schematic flow chart of a variable sliding window decoding method for double-space coupling LDPC codes according to the present invention;

FIG. 2 is a schematic diagram of a coupling process of a (3, 6, L) SC-LDPC code;

FIG. 3 is a diagram of a prototype DSC-LDPC code for a joint source-channel coding system;

FIG. 4 is a decoding window of SWD sliding along a master pattern of a DSC-LDPC code;

FIG. 5 is a diagram illustrating the expansion process of the decoding window in the parity check matrix according to the present invention;

FIG. 6 is a comparison of decoding performance for the present invention;

FIG. 7 is a schematic diagram of a variable sliding window decoding system of a double-spatial coupling LDPC code according to the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The invention aims to provide a variable sliding window decoding method and system of a double-space coupling LDPC code, which improve the decoding performance.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

Example 1

The invention discloses a variable sliding window decoding method of a double-space coupling LDPC code.

The source generates a random sequence obeying a Bernoulli distribution, the probability p of a bit being 1 in said random sequence ₁ Is 0.02.

<xnotran> , (1,0,1,0,0,0,0,0,0,0). </xnotran>

It can be seen from fig. 3 that there are 4 variable nodes and 1 check node in a source protograph unit, and thus

Represents the number of variable nodes in the source proto-pattern unit, and is used for judging whether the number of the variable nodes is greater than or equal to the value of the variable nodes in the source proto-pattern unit>

Representing the number of check nodes in the source protograph cell.

Based on the characteristics of the convolution code of the SC-LDPC code, a partial checksum coding algorithm is adopted in a signal source and a signal channel, and the method specifically comprises the following steps:

and carrying out compression coding on the random sequence to obtain a sequence after source coding.

And coding the sequence after the information source coding based on the information bit and the check bit on the channel to obtain the sequence after the channel coding.

Source coding: source code word s _[0,L-1] Can use a check matrix

Compression-encode it into u _[0,L-1] ，

The method is a formula of source coding, and the source coding is to carry out compression coding on a random source sequence, so that redundant information is reduced, and the reliability of a communication system is improved. The output of the source encoder would be the input of the channel encoder. <xnotran> (1,0,1,0,0,0,0,0,0,0) u </xnotran> _[0,L-1] ＝(0,1,1,0,0)。

Channel coding: in order to satisfy source and channel matching, the source coupling length and the channel coupling length satisfy L ^sc +m _s ＝L ^cc -m _s . It can be seen from fig. 3 that there are 2 variable nodes and 1 check node in a channel prototype graph unit, so that

Representing the number of variable nodes in a channel protogram cell, in combination with a channel profile>

The representation represents the number of check nodes in the channel protogram cell.

The variable nodes of the channel connected to the source check nodes are called information bits, e.g., (0, 1, 0) (nodes hatched in FIG. 3), and the information bits are encoded by

The variable nodes not connected to the source check node in the same bitmap as the information bits are also called check bits, for example, the check bits are (0, 1, 0) (fig. 3 is a node without padding). The coding of the check bit is->

Wherein k denotes the number of iterations and j denotes the jth variable node, </or >>

Indicating the ^ th or greater in the check matrix>

Element(s) is/are present>

Indicating the ^ th or greater in the check matrix>

Each element, i, represents the ith primitive unit. The variable nodes left added additionally for source and channel matching are called deletion nodes, for example, the deletion nodes are (0, 0) (nodes of dotted line in fig. 3); for example, the sequence after channel coding is: (information bit 1 parity bit 1, information bit 2 parity bit 2, information bit 3 parity bit 3,information bit 4 parity bit 4, information bit 5 parity bit 5, delete node 1 delete node 2, delete node 3 delete node 4), the source coded sequence (0, 1, 0) is channel coded to obtain the sequence (00, 11,10, 00).

Information bits are bits that carry information. The check bit is the redundant bit information generated after the information bit is encoded, and is used for error correction and error detection. The node is deleted as an extra node for meeting the matching relation between the source and the channel in the joint source channel, but the node does not carry out coding and decoding, and the complexity of the system is not improved. The output of the channel encoder is modulated and then sent to a channel for transmission, the decoding end receives the information transmitted by the channel, and the receiving end decodes the information to obtain useful information.

And modulating the sequence after the channel coding by adopting BPSK modulation to obtain a modulated signal.

BPSK modulation is to modulate the information with bit 1 to-1 and the information with bit 0 to +1; for example, the channel-coded sequence is (00, 11,10, 00) BPSK modulated to (+ 1, -1-1, -1+1, + 1).

BPSK is binary phase shift keying and is the most basic modulation scheme. BPSK uses the phase change of a carrier to transmit digital information, with the amplitude and frequency remaining unchanged during transmission. The function of the sequence modulation is to reduce interference, improve the anti-interference capability of the system, and simultaneously realize the interchange between transmission bandwidth and signal-to-noise ratio. Modulation may also process information that needs to be transmitted into a form suitable for transmission over a channel. The output of the modulator is fed into a channel, where noise affects the signal, and the decoder removes some of the effects by decoding.

The modulated signal is transmitted to the decoder over an AWGN channel.

The noise of the AWGN channel causes interference to the incoming symbols, and the interference of the noise is random. The signals after being interfered in the above example may be:

(1.52,0.68,2.4,-1.6,-0.02,0.74,1.36,2.56,-2.3,1.3,-2.35,-0.33,+1.52,+1.23)。

the modulated signal is transmitted over an AWGN channel with a channel model represented as y = x + n, where x is the channel input symbol, n is the mean 0, and the variance σ is ² Gaussian noise.

The demodulator demodulates the channel output symbol probabilities into bit Log-Likelihood Ratios (LLRs).

Fig. 1 is a schematic flow chart of a variable sliding window decoding method of a double-space coupling LDPC code of the present invention, and as shown in fig. 1, a variable sliding window decoding method of a double-space coupling LDPC code includes the following steps:

step 101: and initializing the decoding window width of a decoder of the double-space coupling LDPC code.

The window size of the sliding window decoding algorithm of the SC-LDPC code has a range: from the coupling width plus 1 to the coupling length. Once the decoding window size is set, the decoder performs BP decoding within the decoding window. The window size set at this time is 3, so the information that the first window only needs to receive is (1.52, 0.68,2.4, -1.6, -0.02, 0.74).

The maximum value of the preset window is 16, and the initial width of the decoding window is 3.

After the decoder receives the code words, the size of a decoding window is initialized, and BP decoding is carried out on the code words in the decoding window.

The iterative process is executed in the window, the window size is sequentially expanded by adopting the window expansion scheme provided by the invention, and the specific window expansion scheme is as follows:

step 102: BP decoding is carried out on the received code words in a current decoding window of a current decoding position, and the log-likelihood ratio of the target symbol in the current decoding window is obtained and recorded; the current decoding position is the position of the upper left corner of the current decoding window; and the target symbol is the leftmost source original model graph unit of the current decoding window.

The expansion process of the decoding window in the parity check matrix is shown in FIG. 5, b in FIG. 5 _v Representing the number of variable nodes, b _c Indicating the number of check nodes and W the width of the decoding window.

The BP decoding is to decode a signal subjected to noise interference in a received channel, and an LLR value obtained in this step is used for subsequent judgment output, so as to obtain a final result of a decoder.

Initializing LLRs received within an initial decoding window to obtain

And &>

Wherein the content of the first and second substances,

an initialized LLR value representing a source's vth target symbol, <' >>

Initialized LLR value, p, representing the v-th symbol of the channel ₁ Is the probability of a bit of 1 in the source, σ ² Representing the variance of the noise, y _v Indicating that the decoder received the LLR value for the v-th symbol. And the variable node of the information source is the target symbol. />

Obtained on initialisation

And &>

Iterating to obtain>

Wherein the content of the first and second substances,

the LLR information which represents that the variable node v is transmitted to the check node c by the source decoder during the kth decoding iteration; />

The LLR information which represents that the variable node v transmits to the check node c when the channel decoder performs the decoding iteration at the kth time; />

The LLR information transmitted to the variable node v by the check node c is transmitted by the source decoder during the kth decoding iteration; />

The LLR information which is transmitted to the variable node v by the check node c when the channel decoder decodes iteration at the kth time is represented; />

The LLR information transmitted to the channel variable node v by the transmission source check node c in the kth decoding iteration is represented;

and the LLR information transmitted to the source check node c by the transmission channel variable node v at the kth decoding iteration is represented.

Wherein, the first and the second end of the pipe are connected with each other,

formula (2) is to calculate the LLR value transmitted from the variable node to the check node in the source decoder and the channel decoder; equation (3) is to calculate LLR values passed from check nodes to variable nodes in the source decoder and the channel decoder.

Obtained from initialization and iteration

And &>

And (3) decoding judgment:

wherein, L(s) _v ) Indicating that the LLR values are obtained iteratively.

And (3) judging LLR values:

/>

wherein the content of the first and second substances,

is the final decoding result and is compared with the source sequence to estimate whether the decoding is accurate.

Step 103: and calculating the log-likelihood ratio threshold value and the average value of the log-likelihood ratios of the target symbols of all the decoding windows recorded at the current decoding position.

The average of the log-likelihood ratios of the target symbols for all decoding windows recorded at the current decoding position is expressed as:

wherein the content of the first and second substances,

representing the average log-likelihood ratio of the target symbol.

The log-likelihood ratio threshold is expressed as:

where θ represents a log likelihood ratio threshold, L ^sc Representing the coupling length of the source, m _s Indicating the coupling width, W, of the source _max Represents the maximum value of the preset window, W _inc Represents a window increment, W _inc ＝W _max -W _f ，W _f Which represents the initial width of the decoding window,

represents the average log-likelihood ratio of the target symbol and t represents the position of the target symbol.

Step 104: and judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or the width of the current decoding window reaches the maximum value of a preset window.

If not, go to step 105.

Step 105: and adding 1 to the width of the current decoding window, and returning to the step 102.

If yes, go to step 106.

Step 106: and (4) finishing decoding the code words in the current decoding window, sliding the current decoding window to the next decoding position, and returning to the step 102 until all the code words received by the decoder finish decoding.

The sliding the current decoding window to the next decoding position specifically includes: and sliding the current decoding window to the right by one original pattern unit to decode the next source target symbol.

Simulation results of the traditional SWD decoding performance of DSC-LDPC codes of the JSC system and the simulation result are shown in FIG. 6, the decoding performance of the invention is superior to that of the traditional SWD under different initial window sizes, and the simulation result proves the effectiveness of the invention in reducing the decoding performance loss of the DSC-LDPC codes.

As a specific embodiment, the coupling length L of the source ^sc Is 16, coupling width m _s Is 2, the spreading factor M is 160; coupling length L of channel ^cc 20, coupling width m _s Is 2 and the spreading factor M is 160.

In fig. 6, SWD represents decoding performance of the conventional SWD, SVWD represents decoding performance of the present invention, and in fig. 6, the abscissa represents signal-to-noise ratio and the ordinate represents bit error rate.

Example 2

Fig. 7 is a schematic structural diagram of a variable sliding window decoding system of a double-space coupling LDPC code of the present invention, and as shown in fig. 7, a variable sliding window decoding system of a double-space coupling LDPC code includes:

a decoding window width initialization module 201, configured to initialize a decoding window width of a decoder of the double-space coupling LDPC code.

A log-likelihood ratio determining module 202 for the target symbol in the current decoding window, configured to perform BP decoding on the received codeword in the current decoding window at the current decoding position, obtain and record the log-likelihood ratio of the target symbol in the current decoding window; the current decoding position is the position of the upper left corner of the current decoding window; and the target symbol is the leftmost source original model graph unit of the current decoding window.

And the mean value calculation module 203 of the log-likelihood ratio is used for calculating the threshold value of the log-likelihood ratio and the mean value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position.

The determining module 204 is configured to determine whether the average value of the current log-likelihood ratio is greater than a current log-likelihood ratio threshold or whether the width of the current decoding window reaches a preset window maximum value.

A decoding window widening module 205, configured to add 1 to the width of the current decoding window when the determining module determines that the target symbol is not a target symbol in the current decoding window, and return to the log-likelihood ratio determining module 202.

A current decoding window sliding module 206, configured to, when the determining module determines that the decoding is completed, slide the current decoding window to a next decoding position, and return to the log-likelihood ratio determining module 202 for the target symbol in the current decoding window until all the codewords received by the decoder are decoded.

The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the description of the method part.

The principle and the embodiment of the present invention are explained by applying specific examples, and the above description of the embodiments is only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (8)

1. A variable sliding window decoding method of double space coupling LDPC code is characterized by comprising the following steps:

initializing the decoding window width of a decoder of the double-space coupling LDPC code;

BP decoding is carried out on the received code words in a current decoding window of a current decoding position, and the log-likelihood ratio of a target symbol in the current decoding window is obtained and recorded; the current decoding position is the position of the upper left corner of the current decoding window; the target symbol is the leftmost source original model graph unit of the current decoding window;

calculating the threshold value of the log-likelihood ratio and the average value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position;

judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or the width of the current decoding window reaches the maximum value of a preset window;

if not, adding 1 to the width of the current decoding window, and returning to the step of carrying out BP decoding on the received code word in the current decoding window at the current decoding position to obtain the log-likelihood ratio of the target symbol in the current decoding window;

if so, completing decoding of the code word in the current decoding window, sliding the current decoding window to the next decoding position, returning to the step of performing BP decoding on the received code word in the current decoding window at the current decoding position to obtain the log-likelihood ratio of the target symbol in the current decoding window until all the code words received by the decoder are completely decoded.

2. The variable sliding window decoding method of a double spatial coupling LDPC code according to claim 1, wherein before initializing a decoding window width of a decoder of the double spatial coupling LDPC code, the method specifically comprises:

the information source generates a random sequence obeying Bernoulli distribution, and the probability that a bit in the random sequence is 1 is 0.02;

carrying out compression coding on the random sequence to obtain a sequence after source coding;

coding the sequence after the information source coding based on the information bit and the check bit on the channel to obtain the sequence after the channel coding;

adopting BPSK modulation to modulate the sequence after the channel coding to obtain a modulated signal;

the modulated signal is transmitted to the decoder over an AWGN channel.

3. The variable sliding window decoding method of double spatial coupling LDPC code according to claim 2, wherein the coupling length L of the source ^sc 16, coupling width m _s Is 2 and the spreading factor M is 160.

4. The method of claim 3, wherein the coupling length L of the channel is L ^cc 20, coupling width m _s Is 2 and the spreading factor M is 160.

5. The variable sliding window decoding method of double spatial coupling LDPC code according to claim 1, wherein the log-likelihood ratio threshold is expressed as:

where θ represents a log likelihood ratio threshold, L ^sc Representing the coupling length of the source, m _s Indicating the coupling width, W, of the source _max Represents the maximum value of the preset window, W _inc Represents a window increment, W _inc ＝W _max -W _f ，W _f Which represents the initial width of the decoding window,

represents the average log-likelihood ratio of the target symbol and t represents the position of the target symbol.

6. The variable sliding window decoding method of the double spatial coupling LDPC code according to claim 1, wherein the preset window maximum value is 16 and the initial width of the decoding window is 3.

7. The variable sliding-window decoding method of a double-spatial-coupling LDPC code according to claim 1, wherein sliding a current decoding window to a next decoding position specifically comprises:

and sliding the current decoding window to the right by one original image unit.

8. A variable sliding window decoding system for a double spatially coupled LDPC code, comprising:

a decoding window width initialization module for initializing the decoding window width of the decoder of the double space coupling LDPC code;

a log-likelihood ratio determining module of the target symbol in the current decoding window, which is used for carrying out BP decoding on the received code word in the current decoding window at the current decoding position, obtaining the log-likelihood ratio of the target symbol in the current decoding window and recording the log-likelihood ratio; the current decoding position is the position of the upper left corner of the current decoding window; the target symbol is the leftmost source original model graph unit of the current decoding window;

the average value calculation module of the log-likelihood ratio is used for calculating the threshold value of the log-likelihood ratio and the average value of the log-likelihood ratios of the target symbols of all decoding windows recorded at the current decoding position;

the judging module is used for judging whether the average value of the current log-likelihood ratio is larger than the current log-likelihood ratio threshold or whether the width of the current decoding window reaches the maximum value of a preset window;

a decoding window widening module, configured to add 1 to the width of the current decoding window and return to the log-likelihood ratio determining module of the target symbol in the current decoding window when the determining module determines that the target symbol is not in the current decoding window;

and the current decoding window sliding module is used for sliding the current decoding window to the next decoding position when the judgment module judges that the code word in the current decoding window completes decoding, and returning to the log-likelihood ratio determining module of the target symbol in the current decoding window until all the code words received by the decoder complete decoding.

Images