CN112134571A - Sliding window decoding method and device of space coupling LDPC code - Google Patents

Abstract

The invention discloses a sliding window decoding method and a sliding window decoding device of a space coupling LDPC code, wherein the method comprises the following steps: initializing parameters of sliding window decoding according to the coupling length, the smooth parameters, the edge extension patterns and the lifting factors of the space coupling LDPC code and by combining channel conditions and receiving conditions, wherein the parameters of the sliding window decoding comprise window length and the maximum iteration times of each window; and decoding the sliding window according to the window length, the maximum iteration times of each window and the set advanced sliding condition and the advanced termination condition. The method can comprehensively optimize the decoding performance of sliding window decoding, hardware storage resource occupation and decoder delay through dynamic configuration of the window size and the maximum iteration times of the window, and meanwhile, by introducing a sliding condition in advance and a termination condition in advance, the decoder delay can be obviously reduced, so that the operation resources are more efficient, and the decoder throughput capacity is higher.

Description

Sliding window decoding method and device of space coupling LDPC code

Technical Field

The invention relates to the technical field of digital information transmission, in particular to a sliding window decoding method and device of a space coupling LDPC code.

Background

One of the fundamental tasks of digital communication systems is to achieve efficient transmission of digital information, and error control using channel coding is an efficient way to achieve this task. R.g. gallager invented Low-Density Parity-Check (LDPC) codes in the 60's of the 20 th century. In digital communication systems, LDPC codes are widely used for their decoding performance approaching the shannon limit and corresponding low-complexity Belief Propagation (BP) iterative decoding algorithms that are linear in code length.

QC-SC-LDPC codes are LDPC codes with quasi-cyclic and spatial coupling characteristics, and a sliding window decoding method is generally adopted. The conventional sliding window decoding method generally faces the following technical problems in the application occasions of channel condition change or code rate compatible code length extensible:

1) the window size is fixed: in the conventional sliding window decoding method, the window size is generally set to a fixed value. The method has the defect that the window size cannot be dynamically configured according to parameters such as an edge extension pattern, a coupling length L, a smoothing parameter w, a lifting factor Z and the like of the QC-SC-LDPC code, and factors such as a coding algorithm in a window and the limitation of coding complexity, and the like by combining channel conditions and receiving conditions, so that the decoding performance of sliding window decoding, hardware storage resource occupation, decoder delay and the like cannot reach comprehensive optimal values.

2) Intra-window decoding fixing: in the traditional sliding window decoding, the decoding iteration number of each window is fixed as the maximum iteration number I_max. The method has the defects that the sliding of the window cannot be dynamically adjusted according to the decoding condition of variable nodes or check nodes in the window, so that part of the window which can be successfully decoded only by a small number of iterations generates the waste of decoding iteration times, and the decoding of the window which can be successfully decoded by a large number of iterations is insufficient, so that the error transmission phenomenon is generated until the decoding fails. In addition, for the situation that the error transmission exists in the sliding window decoding and the subsequent decoding failure probability is increased sharply, the decoding cannot be stopped in time and the waste of operation resources is generated.

Disclosure of Invention

The present invention is directed to solving, at least to some extent, one of the technical problems in the related art.

Therefore, one objective of the present invention is to provide a sliding window decoding method for spatially-coupled LDPC codes, which can comprehensively optimize decoding performance of sliding window decoding, hardware storage resource occupation, and decoder delay; by introducing the early sliding condition and the early termination condition, the time delay of the decoder can be obviously reduced, the operation resources are more efficient, and the throughput capacity of the decoder is higher.

Another objective of the present invention is to provide a sliding window decoding apparatus for spatially coupled LDPC codes.

To achieve the above object, an embodiment of an aspect of the present invention provides a sliding window decoding method for a spatial coupling LDPC code, including:

s1, initializing parameters of sliding window decoding according to the coupling length, the smooth parameters, the edge extension pattern and the lifting factor of the space coupling LDPC code, and combining the channel condition and the receiving condition, wherein the parameters of the sliding window decoding comprise the window length and the maximum iteration times of each window;

and S2, performing sliding window decoding according to the window length, the maximum iteration times of each window, and the set advanced sliding condition and the advanced termination condition.

In order to achieve the above object, another embodiment of the present invention provides a sliding window decoding apparatus for spatially coupled LDPC codes, including:

a determining module, configured to initialize a sliding window decoding parameter according to a coupling length, a smoothing parameter, an edge extension pattern, and a lifting factor of a spatial coupling LDPC code, in combination with a channel condition and a receiving condition, where the sliding window decoding parameter includes a window length and a maximum number of iterations of each window;

and the decoding module is used for carrying out sliding window decoding according to the window length, the maximum iteration times of each window and a set advanced sliding condition and an advanced termination condition.

The sliding window decoding method and device of the space coupling LDPC code have the following beneficial effects:

(1) dynamically configuring the size of a window in a maximum window range according to factors such as basic parameters of an SC-LDPC code, an intra-window decoding algorithm, decoding complexity limitation, channel conditions, receiving conditions and the like, and configuring the maximum iteration times of the window at the same time so as to comprehensively optimize the decoding performance of sliding window decoding, hardware storage resource occupation and decoder delay;

(2) by introducing the early sliding condition and the early termination condition, the window sliding and the decoding termination of the sliding window decoding can be adaptively controlled, so that the decoding complexity in high signal-to-noise ratio and low signal-to-noise ratio is effectively reduced, and the signal-to-noise ratio threshold is not deteriorated. In practical application, the time delay of the decoder can be obviously reduced, so that the operation resources are more efficient, and the throughput capacity of the decoder is higher.

Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.

Drawings

The foregoing and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

FIG. 1 is a flow chart of a sliding window decoding method of a spatially coupled LDPC code according to an embodiment of the present invention;

FIG. 2 is a diagram illustrating a QC-SC-LDPC code-based matrix and a window according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of a QC-SC-LDPC code-base matrix according to an embodiment of the present invention;

FIG. 4 is a diagram of N according to one embodiment of the present invention_wWhen the signal-to-noise ratio difference of the QC-SC-LDPC codes is 9-20, analyzing the result of polygonal RCA (Rac code analysis) of the relation between the signal-to-noise ratio difference and the decoding complexity;

FIG. 5 is a diagram of N according to one embodiment of the present invention_wWhen the QC-SC-LDPC code is 12-17, obtaining an actual simulation result of the relationship between the signal-to-noise ratio difference and the decoding complexity of the QC-SC-LDPC code;

FIG. 6 is a flow chart of the kth window decoding according to one embodiment of the present invention;

FIG. 7 is a diagram illustrating simulation results of sliding window decoding according to an embodiment of the present invention;

FIG. 8 is a block diagram of an apparatus for sliding window decoding of spatially coupled LDPC codes according to an embodiment of the present invention.

Detailed Description

Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.

An LDPC code may be represented by a check matrix H or an equivalent bipartite graph (called a Tanner graph). In the check matrix representation, the codeword space is the nulling space of the check matrix H. In the Tanner graph representation, the check node corresponds to a row of H, the variable node corresponds to a column of H, and the non-zero element of H corresponds to an edge connecting the variable node and the check node.

A space coupling (SC-) LDPC code is a special novel LDPC code composed of L non-coupling LDPC subcodes H_unc,0,…,_unc,L-1Obtained by spatial coupling, where L is called coupling length. The spatial coupling of the subcodes can be obtained through edge spreading (edge spreading) operation, and the process of edge spreading, namely variable nodes between adjacent subcodes, form a check relation through the spread edges. Check matrix H of SC-LDPC code by edge extension_SCThe following can be defined:

wherein the content of the first and second substances,

called subcode H_unc,iThe edge extension pattern w is called a smoothing parameter (smoothing parameter), i.e. the maximum number of sub-codes sharing a check node with a certain sub-code through the edge extension process. To reduce the description complexity and hardware implementation complexity, the L sub-codes and their edge spreading patterns are usually designed to be the same (i.e. they are designed to be identical)

H_unc,i＝H_unc,H_pattern,i＝H_pattern＝[H₀ ^T,H₁ ^T,…,H_w ^T]^TI is more than or equal to 0 and less than or equal to L-1), at the moment, the check matrix H of the SC-LDPC code_SCCan be expressed as:

wherein, the non-coupled subcode check matrix H_uncIs of size M_unc×N_uncThe process of edge extension, is easy to know

In addition to the above design, another possible design is that the spreading pattern varies periodically every T sub-codes and their edges, i.e. H_unc,i＝H_unc,T+iAnd is and

wherein i is more than or equal to 0 and less than or equal to L-T-1, and i is more than or equal to 0₀Less than or equal to T-1, and i₀≡ i (mod T), uncoupled subcode check matrix H_unc,iAll are M_unc×N_unc。

The Quasi-cyclic structure of Quasi-cyclic (QC-) LDPC code is a traditional LDPC code structure with simple design, convenient description and flexible realization, and the structure is also commonly used for SC-LDPC code to obtain QC-SC-LDPC code which is convenient for description and hardware realization. Check matrix H of QC-SC-LDPC code_SCCan be composed of smaller base matrix B_SCThe code is obtained through a lifting process, and the basic matrix B of the SC-LDPC code is based on that L sub-codes and edge expansion patterns thereof are the same or that each T sub-codes and edge expansion patterns thereof periodically change_SCThe structures can all be represented as follows:

that is, the base matrix spreading pattern of each sub-code is the same, except that the lifting matrices of the L sub-codes are either the same, or the lifting matrices of each T sub-codes vary periodically. Lifting procedure is to base matrix B_SCIs non-zeroReplacing elements with one cyclic shift sub-matrix (or permutation sub-matrix) or superposition of a plurality of cyclic shift sub-matrices (or permutation sub-matrices) with the size of Z multiplied by Z, and combining the basic matrix B_SCIs replaced by a zero matrix of size Z × Z, Z being called the lifting factor. Non-coupled subcode corresponding base matrix B_uncIs m in size_unc×n_uncWherein M is_unc＝m_uncZ，N_unc＝n_uncZ。

Code rate R of sub-code_trueCalled True code Rate (True Rate), R_true＝(n_unc-m_unc)/n_uncHere, it is assumed that no variable node is deleted. Code rate R of QC-SC-LDPC code_designCalled Design code Rate (Design Rate), R_design＝(Ln_unc-(L+w)m_unc)/Ln_uncDue to the two-end termination characteristic of space coupling of QC-SC-LDPC codes, the designed code rate has inevitable code rate loss delta R (equal to wm) compared with the real code rate_unc/Ln_unc. One way to reduce the rate loss is to remove the last Δ w rows of the QC-SC-LDPC code basis matrix by a row truncation operation, and the rate loss can be reduced to Δ R ═ wm_unc-Δw)/Ln_unc. After the truncation operation, the base matrix size finally becomes m × n ═ m ((L + w) m_unc-Δw)×Ln_uncAccordingly, the check matrix size of the QC-SC-LDPC code becomes mxn ═ mzxnz.

For the convenience of parallel decoding and hardware implementation and the description of the decoding method of the invention, each Z row of the check matrix is defined as one layer, the QC-SC-LDPC code comprises 1 st layer to m th layer, and the i (1 ≦ i ≦ m) layer corresponds to the base matrix B_SCRow i of (2).

SC-LDPC codes have proven to have a common property in an asymptotic sense, i.e., the ability to approach channel capacity in almost all channels. Therefore, SC-LDPC codes and their coded modulation schemes have a stronger potential to cope with flexible and variable scenarios in digital communication systems than conventional LDPC codes and their coded modulation schemes optimized for specific channel conditions or receiver types.

One Decoding method of the SC-LDPC code is Full Block Decoding (Full Block Decoding), that is, DecodingCombining SC-LDPC code H_SCDecoding is performed as a conventional LDPC block code. However, SC-LDPC codes designed to approximate channel capacity and maintain common characteristics typically have a large coupling length L and a long subcode length N_uncTotal code length N ═ LN_uncGenerally, the decoding method is large, and therefore, in practical application, the whole block decoding method faces the problems that hardware storage resources occupy large, the decoder has high delay, the operation resources are low in efficiency, the throughput capacity is limited, and the like.

Another Decoding method of the SC-LDPC code is Sliding Window Decoding (Sliding Window Decoding). The decoding method firstly defines the sequence number as-W₁，-W₁+1，...，W₀-1，W₀W ═ W₀+W ₁1 window, wherein each window corresponds to the SC-LDPC code check matrix H_SCA sub-graph of the corresponding Tanner graph. The conventional sliding window decoding process is: sequentially carrying out decoding iteration on the kth window cycle, and if the kth window is not the last window, sliding to the (k + 1) th window until the maximum iteration times are met; if the k window is the last window and the normal termination condition is satisfied, the decoding is terminated. The invention introduces the operation of sliding in advance and terminating in advance, the improved sliding window decoding process is: sequentially carrying out decoding iteration on the kth window in a circulating manner, and if the kth window is not the last window, sliding to the (k + 1) th window until a set reliability condition is met; if the kth window is the last window and meets the normal termination condition, or the kth window is not the last window and meets the early termination condition, the decoding is terminated. The reliability condition of the non-last window is a sliding condition, and comprises a normal sliding condition when the window decoding iteration frequency reaches the maximum iteration frequency, or an advanced sliding condition when the window decoding iteration frequency does not reach the maximum iteration frequency; wherein, according to the actual requirement and the irregular structure characteristic of the SC-LDPC code, the last window is not necessarily the W-th window₀A window.

In a common window definition, each window is defined as a plurality of check nodes and variable nodes associated with the check nodes, which are called as check nodes in a centralized and continuous (CN-centered) manner) The window of (2). Each window is typically associated with the same number of check nodes, except for windows that pass through the start or end row. For window definition of QC-SC-LDPC codes, each window contains N consecutive N, except for a window containing a start layer or an end layer_wLayer of which N_wThe window length defined for the present invention; meanwhile, the layer number of the first layer of the window sequentially increases with the window number (e.g., in a typical definition, the kth window includes the kth, k +1, …, k + N_w-1 layer).

Compared with a whole-block decoding method, under the window definition, each decoding iteration of sliding window decoding is only carried out aiming at one window, so that the storage space randomly read by hardware occupies smaller resources; meanwhile, the sliding window decoder can input the soft information of the coded bits and output the decoded code words in a 'stream' form, and the delay of the decoder is obviously reduced; meanwhile, by combining layer scheduling decoding of external information rapid transmission, the computing resources are efficient, and the throughput capacity is improved. In addition, under the condition of reasonable decoding parameter selection, compared with a whole-block decoding method, the sliding window decoding method does not generate obvious decoding performance loss.

The following describes a sliding window decoding method and apparatus for a spatially coupled LDPC code according to an embodiment of the present invention with reference to the drawings.

First, a sliding window decoding method of a spatially coupled LDPC code proposed according to an embodiment of the present invention will be described with reference to the accompanying drawings.

FIG. 1 is a flowchart of a sliding window decoding method of a spatially coupled LDPC code according to an embodiment of the present invention.

As shown in fig. 1, the sliding window decoding method of the spatial coupling LDPC code includes the following steps:

step S1, initializing the parameters of sliding window decoding according to the coupling length, the smooth parameter, the edge extension pattern and the lifting factor of the space coupling LDPC code, combining the channel condition and the receiving condition, wherein the parameters of sliding window decoding comprise the window length and the maximum iteration times of each window.

Specifically, the channel strips may be combined according to the coupling length L, the smoothing parameter w, the edge extension pattern, the lifting factor Z, etc. of the QC-SC-LDPC codeCondition and receiving condition, initializing parameters of sliding window decoding, the parameters of sliding window decoding including window length N_wMaximum number of iterations per window I_maxAnd the like.

The definitions of the QC-SC-LDPC code, the coupling length L, the smoothing parameter w, the edge extension pattern, the lifting factor Z, and other parameters are described in detail above, and are not described herein again.

As shown in FIG. 2, the window employed in one implementation is defined as a continuous (CN-centered) window of the check node set, where a row of the base matrix corresponds to Z check nodes of the check matrix. All window lengths are fixed values of N, except for the windows passing through the start and end layers_w. Specifically, at window length N_wNext, the kth window includes the kth, k +1_w-1 layer, wherein k ═ 2-N_w，3-N_w,., m-1, m, parameter W₁＝N_w-2，W₀M, W₀+W₁-1＝m+N_w-1 window. The window passing through the starting layer and the ending layer needs to consider the boundary condition of window sliding, namely only the 1 st layer to the m th layer in the window are effective. E.g. k-2-N_wThe time window should include 2-N_w，3-N_w,., layer 0, 1, but only layer 1 is within the check matrix of the QC-SC-LDPC code, so the window actually contains only layer 1.

Definition of window decoding as described above, in particular, in the embodiment, decoding of each window preferably employs Sum-Product decoding Algorithm (SPA) which specifies row-column operation of LDPC decoding and layer-scheduled decoding Algorithm (layer-scheduled decoding) which specifies operation sequence of row-column operation units of LDPC decoding.

Specifically, the k-th window actually contains N layers_w，k＝min(k+N _w1, m) -max (1, k) +1, the process of one decoding iteration for the k-th window is to perform the 1 st, 2_w，kAnd decoding layer by layer. After the layer-by-layer decoding, i.e. after a certain layer in the window completes one SPA decoding line operation, the bit soft information of the layer of associated variable node is immediately updated (i.e. the associated variable is subjected toThe quantum node performs a partial column operation) and is used for decoding of the next layer. Further, alternative or simplified algorithms to sum-product algorithms, such as various min-sum decoding algorithms, may also be applied to the sliding window decoding method proposed in the embodiments of the present invention. Further, the layer-by-layer decoding sequence of one decoding iteration process in the layer decoding scheduling algorithm can be adjusted according to needs.

Window length N_wAt less than the maximum window length N_w，maxConfigurable within range, maximum number of iterations per window I_maxIt may also be configured, determined by the limitations of the actual decoding complexity (or hardware throughput capability).

In a specific embodiment, the window decoding complexity is defined as the arithmetic mean of the total number of row operations in all windows per row of the SC-LDPC code. Setting the iteration times of each window to be fixed and reaching the maximum iteration times I_maxThen, the window length is N_wWhile, the window decoding complexity is fixed to N_wI_max。

In a particular embodiment, the window length N_wWhen the number of window iterations is fixed, the decoding performance is optimal under the same window decoding complexity, namely the window length with the minimum distance (called signal-to-noise ratio gap) between the decoding signal-to-noise ratio threshold and the shannon limit under the limitation of a typical channel (such as a BI-AWGN channel) and a given bit error rate (or block error rate). Preferably, a multilateral type Density Evolution (MET-DE) or multilateral type Approximation to easy Channel (RCA) asymptotic analysis tool is used to analyze the window length with the optimal decoding performance when the lifting factor Z approaches infinity. Further, the finite code length effect and the non-ideal factors of the actual SC-LDPC code are comprehensively considered, and the window length with the optimal decoding performance when the code length is finite is determined and used as the finally selected window length N_w。

In a specific embodiment, a sum-product decoding algorithm and a layer scheduling decoding algorithm are adopted, and when the number of window iterations is fixed, the window decoding complexity is generally selected to be N_wI_max＝30～60。

In one embodiment, the basis matrix of the QC-SC-LDPC code is referenced in FIG. 3, where the blocksAnd crosses represent elements 1 and 0, respectively. Wherein the uncoupled subcode corresponds to the size m of the base matrix_unc×n _unc1 × 3, the coupling length L is 40, the smoothing parameter w is 6, and the base matrix determines the edge extension pattern as shown in the right part of fig. 3. The base matrix is row truncated, i.e. after Δ w is 4 rows deleted, so the base matrix size m × n ((L + w) m)_unc-Δw)×Ln_unc42 × 120. The lifting factor Z of the base matrix is 768, so the final code length N nZ 92160.

In an exemplary embodiment, as shown in FIG. 4, FIG. 4 illustrates QC-SC-LDPC codes using different window lengths N under a BI-AWGN channel_wAnd performing multi-edge RCA analysis to obtain the relationship between the signal-to-noise ratio difference and the window decoding complexity. Wherein the threshold BER is 10-¹⁰. It can be seen that when the number of window iterations is fixed, the window decoding complexity is selected to be N_wI_maxWhen an SPA decoding algorithm and a layer scheduling decoding algorithm are adopted and the lifting factor Z tends to be infinite, N is 30-60_w13-14 are the window length with the best decoding performance.

In an exemplary embodiment, as shown in FIG. 5, FIG. 5 illustrates QC-SC-LDPC codes using different window lengths N under BI-AWGN channels_wAnd then, carrying out actual sliding window decoding simulation (the window iteration times are fixed, and the advanced sliding and the advanced termination are not adopted), and obtaining the relation between the signal-to-noise ratio difference and the window decoding complexity. Wherein, the BER threshold is 10^-5. It can be seen that when the number of window iterations is fixed, the window decoding complexity is selected to be N_wI_maxAnd (3) adopting an SPA decoding algorithm and a layer scheduling decoding algorithm, and carrying out N-time matching on the constructed QC-SC-LDPC codes when the lifting factor Z is 768_w16 is the window length for which the decoding performance is optimal. Therefore, the finite code length effect and the non-ideal factors of the actual SC-LDPC code may cause the optimal window length to be shifted when the code length is finite. Meanwhile, it can be seen that the window length can be slightly adjusted as needed near the optimal window length due to the robustness of the decoding performance to window length variations.

In a specific embodiment, the window length N is selected by combining the multi-edge RCA analysis and the actual simulation result_w＝16。Binding to N_wI_maxWith a constraint of 30-60, N is selected in the subsequent sliding window decoding implementation_w＝16，I_maxWith 2 being the preferred mode.

It should be noted that the above configuration window length N_wIn the specific implementation process of (3), when the multilateral RCA analysis and the actual simulation experiment are performed, the last window is set as the k-th window to the m-th window. However, in actual sliding window decoding (including conventional sliding window decoding with a fixed number of window decoding iterations), the last window does not need to be the k-th window or the m-th window according to actual needs and the structural irregular characteristic of the QC-SC-LDPC code. In a specific implementation, the last window is generally set as the k ═ m-dW (dW ≧ 0) window, where the value of dW is determined according to the basic parameters (including coupling length L, smoothing parameter w, edge extension pattern, lifting factor Z, etc.) of the QC-SC-LDPC code, and the selected window length N is set as the window length N_wAnd the transmission scheme of the QC-SC-LDPC code and the like.

Step S2, sliding window decoding is performed according to the window length, the maximum number of iterations of each window, and the set advanced sliding condition and advanced termination condition.

S2 further includes:

according to window length N_wAnd the maximum number of iterations I for each window_maxTo k 2-N_wSequentially decoding the k-th window m-dW, and when the k-th window meets the reliability condition, decoding the k-th window by the (2-N)_wK is more than or equal to k and less than m-dW) windows slide to the k +1 th window, and the reliability condition is as follows: the iteration number of the current window decoding reaches the maximum iteration number I_maxPerforming normal sliding, or the current window decoding iteration number does not reach the maximum iteration number I_maxBut the preset advanced sliding condition is met, k is the window serial number of the positive integer, m is the number of rows of the QC-SC-LDPC code base matrix of the positive integer, and dW is the last window parameter of the non-negative integer.

When the kth window meets the decoding termination condition, terminating the decoding, wherein the decoding termination condition is as follows: normal termination condition when kth window is last window, where k is m-dW, or early termination when kth window is not last windowCondition(s) at this time, 2-N_w≤k＜m-dW。

Specifically, the window length N obtained in S1_wWith maximum number of iterations I per window_maxTo k 2-N in sequence_wAnd sequentially decoding the k-th window to the m-dW window.

Wherein, the k (2-N) th_wK is less than or equal to k and m-dW) windows slide to the (k + 1) th window to meet the set reliability condition, including that the window decoding iteration frequency I reaches the maximum iteration frequency I_maxNormal sliding condition of (2), or window decoding iteration number I not reaching maximum iteration number I_maxAdvanced slip conditions in time; wherein the condition for terminating the decoding in the kth window includes a normal termination condition when the kth window is the last window (i.e., k ═ m-dW), or the kth window is not the last window (i.e., 2-N)_wK is less than or equal to m-dW).

In one embodiment of the present invention, as shown in fig. 6, the step of decoding the kth window is as follows:

and S21, receiving the coded bit soft information of the kth window, and performing decoding iteration on the kth window once by adopting a sum-product decoding algorithm (SPA) and a layer scheduling decoding algorithm, wherein the total decoding iteration number of the kth window after iteration is I. After the iteration, a decoding result (namely a hard decision result of each variable node of the QC-SC-LDPC code) and a layer check result (namely the satisfaction condition of each layer check equation of the QC-SC-LDPC code according to the hard decision result) are obtained.

And updating the successful check times of each layer of the QC-SC-LDPC code. Wherein, the layer verification success indicates that each row in the layer is verified successfully. E.g. the i-th layer continuous check before the iteration is successful t_i0If the verification of the ith layer is successful after the iteration, the verification success times of the ith continuous layer are updated to t_i0And +1, otherwise, resetting the verification success times of the ith layer continuous layer.

And recording the check failure proportion of each layer of the QC-SC-LDPC code. The ratio of layer check failure is the ratio of the number of rows of layer check failure to the total number of rows Z.

S22 if I ═ I_max，I_maxFor maximum number of iterations, jumpStep S25, otherwise, proceed to step S23.

S23, if the 1 st, 2 nd in the kth window, the times of the continuous verification of the l layer are not less than t respectively₁，t₂，...，t_lJumping to step S25, otherwise, continuing to step S24; wherein l is the layer number of a positive integer, t₁，t₂，...，t_lThe number of successful continuous checks is a positive integer.

S24, if the k-S +1, k-S +2 of QC-SC-LDPC code, the check failure ratio of k layers (namely the first layer and the previous S-1 layer in the k window) is not less than p₁，p₂，...，p_sJumping to step S26, otherwise returning to step S21, where S is the number of layers of positive integer, p₁，p₂，...，p_sThe check failure ratio is a real number between 0 and 1.

S25, if k is m-dW, i.e. the last window is reached, go to step S26, otherwise, go to the (k + 1) th window.

S26, terminating (or ending) the decoding.

In particular, in the above-described embodiments, sliding window decoding of spatially coupled LDPC codes may be implemented. In order to simplify the control logic of the hardware, l, s, t are set in the specific embodiment_j(1≤j≤l)，p_qAnd (q is more than or equal to 1 and less than or equal to s) are independent of the window serial number k. In addition, the boundary conditions of the windows, i.e. only the effective layers in the determination conditions, still need to be considered in steps S23 and S24. For example, in step S23, if k ≧ 1 and k-l +1 > m, the determination condition in step S25 is changed to 1 st, 2 nd,. m-k +1 th layer in the kth window, and the number of times of successful continuous verification is not less than t₁，t₂，...，t_m-k+1。

The judgment conditions corresponding to steps S22 to S24 do not distinguish between the non-last window and the last window. On the basis of steps S22 to S24, further, the last window may be specially processed according to actual needs. For example, the determination condition corresponding to the case where k is m-dW in step S22 is represented by I_maxModified to be I ═ I'_max(ii) a For example, the determination condition corresponding to k-m-dW in step S23 is modified as appropriate to determine whether all the subcodes are presentAre decoded successfully; for example, when k is m-dW, the advance termination condition corresponding to step S24 may be skipped. In addition, for the non-last window, on the basis of the early termination condition of step S24, a new early termination condition distinguished from step S24 may be further added to terminate decoding early when all the sub-codes are decoded successfully.

In addition to the embodiment of step S1, in an embodiment, dW is 2, l is 3, and t is taken₁，t₂，t₃＝3，3，2，s＝3，p₁＝p₂＝p₃＝0.0625。

FIG. 7 shows the actual simulation results of sliding window decoding according to an embodiment under the conditions of BI-AWGN channel and different SNR. Wherein ET1 considers only the advanced sliding condition in the embodiment (i.e. neglecting step S24, directly returning to step S21), ET2 considers the advanced sliding condition and the advanced termination condition in the embodiment together, and Fixed does not consider the advanced sliding condition and the advanced termination condition in the embodiment (i.e. neglecting steps S23 and S24, each window reaches a Fixed maximum number of iterations I_maxThen slide to the next window) and the last window is the k ═ m windows. FIG. 7(1) is a plot of the ratio of the decoding complexity under ET1 and ET2 to that under Fixed versus the SNR for the two conditions ET1 and ET2, respectively; fig. 7(2) is a plot of maximum sub-code block error rate versus signal-to-noise ratio under three conditions of ET1, ET2, and Fixed. It can be seen that the decoding complexity under the high signal-to-noise ratio can be significantly reduced by the advanced sliding condition in step S23 (for example, the signal-to-noise ratio is 2.85dB, and after the advanced sliding condition is added, the decoding complexity is about 0.56 times that when the window iteration number is fixed); step S24 early termination condition can significantly reduce the decoding complexity at low signal-to-noise ratio (for example, the signal-to-noise ratio is 2.6dB, and after the early termination condition is added, the decoding complexity becomes about 0.57 times when the window iteration number is fixed); under the same threshold of the block error rate, after adding the advanced sliding condition and the advanced termination condition, the deterioration of the threshold of the signal-to-noise ratio does not exceed 0.025 dB.

It should be noted that the above and fixed number of iterations I_maxWhen comparing Fixed instances of (c), an optimal window length of N has been assumed_wThe conditions of (1). Such asCompared with the traditional non-optimal window setting and fixed iteration times I_maxThe SC-LDPC window decoding method provided by the embodiment of the invention has more obvious complexity or performance advantages.

According to the sliding window decoding method of the spatial coupling LDPC code provided by the embodiment of the invention, the window length is configured within the range smaller than the maximum window according to the coupling length, the smooth parameter, the edge extension pattern and the like of the SC-LDPC code, and meanwhile, the condition of sliding ahead and the condition of terminating ahead are introduced in the sliding window decoding process. Therefore, the window length with the optimal decoding performance under the limited decoding complexity can be selected, the decoding performance optimization is facilitated, and the occupation of hardware storage resources can be reasonably arranged; the early sliding condition and the early termination condition can effectively reduce the decoding complexity in high signal-to-noise ratio and low signal-to-noise ratio, and meanwhile, the signal-to-noise ratio threshold is not deteriorated. In practical application, the decoder delay can be obviously reduced, the operation resources are more efficient, and the decoder throughput capacity is higher.

Next, a sliding window decoding apparatus of a spatially coupled LDPC code proposed according to an embodiment of the present invention is described with reference to the drawings.

FIG. 8 is a block diagram of an apparatus for sliding window decoding of spatially coupled LDPC codes according to an embodiment of the present invention.

As shown in fig. 8, the sliding window decoding apparatus of the spatially coupled LDPC code includes: a determination module 801 and a decoding module 802.

The determining module 801 is configured to initialize a sliding window decoding parameter according to the coupling length, the smoothing parameter, the edge extension pattern, and the lifting factor of the spatial coupling LDPC code, in combination with the channel condition and the receiving condition, where the sliding window decoding parameter includes a window length and a maximum number of iterations of each window.

The decoding module 802 is configured to perform sliding window decoding according to the window length, the maximum iteration number of each window, and a set early sliding condition and an early termination condition.

Further, in one embodiment of the present invention, the algorithm of window decoding includes: a sum-product decoding algorithm and a layer scheduling decoding algorithm.

Further, in one embodiment of the present invention, the maximum number of iterations for each window is adjusted according to an actual decoding complexity or hardware throughput, wherein the actual decoding complexity is an arithmetic average of a total number of row operations in all windows for each row of the spatially coupled LDPC code.

Further, in an embodiment of the invention, the decoding module is further configured to decode the window according to the window length N_wAnd the maximum number of iterations I for each window_maxTo k 2-N_wSequentially decoding the k-th window m-dW, and when the k-th window meets the reliability condition, decoding the k-th window by the (2-N)_wK is more than or equal to k and less than m-dW) windows slide to the k +1 th window, and the reliability condition is as follows: the iteration number of the current window decoding reaches the maximum iteration number I_maxPerforming normal sliding, or the current window decoding iteration number does not reach the maximum iteration number I_maxBut meets the set advanced sliding condition; k is the window serial number of the positive integer, m is the number of rows of the QC-SC-LDPC code base matrix of the positive integer, and dW is the last window parameter of the non-negative integer.

When the kth window meets the decoding termination condition, terminating the decoding, wherein the decoding termination condition is as follows: normal termination condition when kth window is last window, where k is m-dW, or early termination condition when kth window is not last window, where 2-N_w≤k＜m-dW。

Further, in an embodiment of the present invention, the decoding module further includes:

the first decoding unit is used for receiving the coded bit soft information of a kth window, performing decoding iteration on the kth window once by adopting a sum-product decoding algorithm and a layer scheduling decoding algorithm, wherein the total decoding iteration frequency of the kth window after iteration is I, obtaining a decoding result and a layer check result after the iteration, updating the successful check frequency of each continuous layer of the space coupling LDPC code, wherein the successful check of the layer indicates that each row in the layer succeeds, and recording the failed check proportion of each layer of the space coupling LDPC code;

a first judging unit for judging if I ═ I_max，I_maxJumping to a second judgment module for the maximum iteration times,otherwise, continuing to perform the advanced sliding unit;

an advance sliding unit for respectively setting the number of times of successful continuous verification of the 1 st layer and the 2 nd layer in the kth window to be not less than t₁，t₂，...，t_lIf not, the system continues to carry out the early termination unit; wherein l is the layer number of a positive integer, t₁，t₂，...，t_lThe number of successful continuous verification times of the positive integer;

a termination unit in advance, which is used for determining if the k-s +1, k-s +2 of the space coupling LDPC code is in k layers, namely the first layer and the previous s-1 layer in the k window, and the proportion of the check failure is not less than p respectively₁，p₂，...，p_sJumping to step S26, otherwise, returning to the first decoding unit; where s is the number of layers of a positive integer, p₁，p₂，...，p_sA check failure ratio of a real number between 0 and 1;

a second judging module, configured to skip to the termination decoding unit if k is m-dW, that is, the last window is reached, and otherwise, slide to the (k + 1) th window;

and a termination decoding unit for terminating the decoding.

It should be noted that the foregoing explanation of the embodiment of the sliding window decoding method for the spatial coupling LDPC code is also applicable to the apparatus of this embodiment, and is not repeated here.

According to the sliding window decoding device of the spatial coupling LDPC code provided by the embodiment of the invention, the window length is configured within the range smaller than the maximum window according to the coupling length, the smooth parameter, the edge extension pattern and the like of the SC-LDPC code, and meanwhile, the early sliding condition and the early termination condition are introduced in the sliding window decoding process. Therefore, the window length with the optimal decoding performance under the limited decoding complexity can be selected, the decoding performance optimization is facilitated, and the occupation of hardware storage resources can be reasonably arranged; the early sliding condition and the early termination condition can effectively reduce the decoding complexity in high signal-to-noise ratio and low signal-to-noise ratio, and meanwhile, the signal-to-noise ratio threshold is not deteriorated. In practical application, the decoder delay can be obviously reduced, the operation resources are more efficient, and the decoder throughput capacity is higher.

Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.

Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.

Claims (10)

1. A sliding window decoding method of a spatial coupling LDPC code is characterized by comprising the following steps:

s1, initializing parameters of sliding window decoding according to the coupling length, the smooth parameters, the edge extension pattern and the lifting factor of the space coupling LDPC code, and combining the channel condition and the receiving condition, wherein the parameters of the sliding window decoding comprise the window length and the maximum iteration times of each window;

and S2, performing sliding window decoding according to the window length, the maximum iteration times of each window, and the set advanced sliding condition and the advanced termination condition.

2. The method of claim 1, wherein the algorithm of window coding comprises: a sum-product decoding algorithm and a layer scheduling decoding algorithm.

3. The method of claim 1, wherein the maximum number of iterations for each window is adjusted according to an actual decoding complexity or hardware throughput, wherein the actual decoding complexity is an arithmetic mean of a total number of row operations in all windows for each row of the spatially coupled LDPC code.

4. The method according to claim 1, wherein the S2 further comprises:

according to the window length N_wAnd the maximum number of iterations I of each of said windows_maxTo k 2-N_wSequentially decoding the k-th window m-dW, and when the k-th window meets the reliability condition, decoding the k-th window by the (2-N)_wK is more than or equal to k and less than m-dW windows slide to the k +1 th window, and the reliability condition is as follows: the current window decoding iteration number reaches the maximum iteration number I_maxPerforming normal sliding, or the current window decoding iteration number does not reach the maximum iteration number I_maxBut the method meets the set advanced sliding condition, wherein k is the window serial number of a positive integer, m is the number of rows of the QC-SC-LDPC code base matrix of the positive integer, and dW is the last window parameter of a non-negative integer;

when the kth window meets a termination coding condition, terminating the coding, wherein the termination coding condition is as follows: normal termination condition when kth window is last window, where k is m-dW, or early termination condition when kth window is not last window, where 2-N_w≤k＜m-dW。

5. The method according to claim 4, wherein the S2 further comprises:

s21, receiving coding bit soft information of a kth window, performing decoding iteration on the kth window once by adopting a sum-product decoding algorithm and a layer scheduling decoding algorithm, wherein the total decoding iteration number of the kth window after iteration is I, obtaining a decoding result and a layer check result after the iteration, updating the number of successful check of each continuous layer of each layer of the space coupling LDPC code, wherein the successful check of the layer indicates that each row in the layer is successfully checked, and recording the proportion of failed check of each layer of the space coupling LDPC code;

s22 if I ═ I_max，I_maxJumping to step S25 for the maximum number of iterations, otherwise continuing to step S23;

s23, if the 1 st, 2 nd in the kth window, the times of the continuous verification of the l layer are not less than t respectively₁，t₂，...，t_lJumping to step S25, otherwise, continuing to step S24; wherein l is the layer number of a positive integer, t₁，t₂，...，t_lThe number of successful continuous verification times of the positive integer;

s24, if the k-S +1, k-S +2 of the space coupling LDPC code, the k layers, namely the first layer and the previous S-1 layer in the k window, the proportion of the check failure is not lower than p respectively₁，p₂，...，p_sJumping to step S26, otherwise returning to step S21; where s is the number of layers of a positive integer, p₁，p₂，...，p_sA check failure ratio of a real number between 0 and 1;

s25, if k is m-dW, that is, the last window is reached, then go to step S26, otherwise, slide to the (k + 1) th window;

and S26, terminating the decoding.

6. A sliding window decoding apparatus for spatially coupled LDPC codes, comprising:

a determining module, configured to initialize a sliding window decoding parameter according to a coupling length, a smoothing parameter, an edge extension pattern, and a lifting factor of a spatial coupling LDPC code, in combination with a channel condition and a receiving condition, where the sliding window decoding parameter includes a window length and a maximum number of iterations of each window;

and the decoding module is used for carrying out sliding window decoding according to the window length, the maximum iteration times of each window and a set advanced sliding condition and an advanced termination condition.

7. The apparatus of claim 6, wherein the window decoding algorithm comprises: a sum-product decoding algorithm and a layer scheduling decoding algorithm.

8. The apparatus of claim 6,

and adjusting the maximum iteration times of each window according to the actual decoding complexity or the hardware throughput capacity, wherein the actual decoding complexity is the arithmetic mean of the total times of row operation of each row of the space coupling LDPC code in all windows.

9. The apparatus of claim 6, wherein the decoding module is further configured to decode the window length N_wAnd the maximum number of iterations I of each of said windows_maxTo k 2-N_wSequentially decoding the k-th window m-dW, and when the k-th window meets the reliability condition, decoding the k-th window by the (2-N)_wK is more than or equal to k and less than m-dW) windows slide to the k +1 th window, and the reliability condition is as follows: the current window decoding iteration number reaches the maximum iteration number I_maxPerforming normal sliding, or the current window decoding iteration number does not reach the maximum iteration number I_maxBut the method meets the set advanced sliding condition, wherein k is the window serial number of a positive integer, m is the number of rows of the QC-SC-LDPC code base matrix of the positive integer, and dW is the last window parameter of a non-negative integer;

when the kth window meets a termination coding condition, terminating the coding, wherein the termination coding condition is as follows: normal termination condition when kth window is last window, where k is m-dW, or early termination condition when kth window is not last window, where 2-N_w≤k＜m-dW。

10. The apparatus of claim 6, wherein the decoding module further comprises:

the first decoding unit is used for receiving the coded bit soft information of a kth window, performing decoding iteration on the kth window once by adopting a sum-product decoding algorithm and a layer scheduling decoding algorithm, wherein the total decoding iteration frequency of the kth window after iteration is I, obtaining a decoding result and a layer check result after the iteration, updating the successful check frequency of each continuous layer of the space coupling LDPC code, wherein the successful check of the layer indicates that each row in the layer succeeds, and recording the failed check proportion of each layer of the space coupling LDPC code;

a first judging unit for judging if I ═ I_max，I_maxJumping to a second judgment module for the maximum iteration times, otherwise continuing to perform an advanced sliding unit;

an advance sliding unit for respectively setting the number of times of successful continuous verification of the 1 st layer and the 2 nd layer in the kth window to be not less than t₁，t₂，...，t_lSkipping to the second judgment module, otherwise continuing to perform an early termination unit; wherein l is the layer number of a positive integer, t₁，t₂，...，t_lThe number of successful continuous verification times of the positive integer;

a termination unit in advance, which is used for determining if the k-s +1, k-s +2 of the space coupling LDPC code is in k layers, namely the first layer and the previous s-1 layer in the k window, and the proportion of the check failure is not less than p respectively₁，p₂,., ps, jumping to step S26, otherwise returning to the first decoding unit; where s is the number of layers of a positive integer, p₁，p₂，...，p_sA check failure ratio of a real number between 0 and 1;

a second judging module, configured to skip to the termination decoding unit if k is m-dW, that is, the last window is reached, and otherwise, slide to the (k + 1) th window;

and a termination decoding unit for terminating the decoding.

Images