CN105577193A - Loop-break based mixed weighted bit-flipping LDPC decoding method - Google Patents

Abstract

The invention discloses a loop-break based mixed weighted bit-flipping LDPC decoding method. The method comprises the following steps of: initializing a decoder and calculating parameters; calculating a check matrix syndrome; calculating a main algorithm decision criterion En1 and flipping a bit z1 most likely wrong; if one bit is flipped repeatedly, calculating an auxiliary algorithm decision criterion En2 again for code words after a hard decision and flipping a bit z2 most likely wrong, and if not, continuing iteration; if one bit is flipped repeatedly again, continuously calculating the main algorithm decision criterion En1 for code word sequences after the processing and flipping the bit z1 most likely wrong, and if not, continuing iteration; and repeating the above steps, and stopping iteration when decoding is successful or a maximum iteration number is reached, and outputting decoded sequences or outputting decoding failure information. The LDPC decoding method provided by the invention has the advantages that the decoding performance is good, the convergence speed of iteration is high, the calculation complexity is low, the realization mode is relatively easy, etc.

Description

Based on the LDPC interpretation method of the mixed weighting bit reversal that loop is eliminated

Technical field

The present invention relates to LDPC decoding field, particularly based on the LDPC interpretation method of the mixed weighting bit reversal of loop elimination.

Background technology

LDPC code, namely low density parity check code (Low-DensityParity-Check) is proposed by Gallager as far back as the sixties in 20th century in its thesis for the doctorate, is a kind of linear block codes based on sparse check matrix definition.Owing to having the excellent properties approaching Shannon limit, and have the attainable encoding and decoding complexity of hardware, structural design, code parameters are selected flexibly.At present, LDPC code has been widely used in the fields such as satellite communication, light communication and deep space communication.LDPC code is adopted by satellite digital video broadcast standard DVB-S2 of future generation, and becomes the strong competitor of forth generation communication system (4G) error correcting coding schemes.

Current LDPC interpretation method is divided three classes: Hard decision decoding, Soft decision decoding and hybrid decoding.Hard decision decoding hardware implementing is simple, and decoding speed is fast, but poor-performing, Hard decision decoding method has bit reversal (bit-flipping, BF) algorithm; Soft decision decoding hardware implementing difficulty, decoding speed is comparatively slow, but performance is best, and soft-decision decoding method has belief propagation (beliefpropagation, BP) algorithm etc.; And namely hybrid decoding is a kind of method adding Soft Inform ation in hard decision, at decoding performance, obtain good compromise between complexity and decoding speed three, therefore can receive much concern.Hybrid decoding method has weighted bit upset (weightedBF, WBF) serial algorithm etc. to represent algorithm.

Weighted bit upset algorithm also experienced by the process of a development, the weighted bit upset algorithm proposed in recent years mainly contains following several: modifiedweightedbit-flipping (M-WBF), lowcomplexityweightedbit-flipping (LC-WBF), reliabilityratiobasedweightedbit-flipping (RR-WBF), fastmodifiedweightedbit-flipping (FM-WBF), improvedmodifiedweightedbit-flipping (IM-WBF) etc.; But above-mentioned algorithm performance still has larger gap compared with soft-decision decoding method, the type of the code that algorithm is suitable for also has larger restriction, and meanwhile, these algorithms are also significantly improved space on decoding speed and complexity.Therefore, better in the urgent need to a kind of performance, applicability is stronger, and convergence rate is faster, the weighted bit method for turning that complexity is not high.

Summary of the invention

The object of the invention is to overcome the shortcoming of prior art and deficiency, a kind of LDPC interpretation method of the mixed weighting bit reversal (Loop-BreakbasedMixedWeightedBit-Flipping, LB-MWBF) based on loop elimination is provided.The method is by the mixing of two existing weighted bit upset algorithms, and reach better than existing weighted bit upset algorithm performance, mean iterative number of time is less, has good effect to length code, Regular codes Irregular codes, does not increase algorithm complex simultaneously.This algorithm is different on the impact of decoding effect in different decoding algorithm based on the loop of check matrix, the method that adopts two kinds of decoding algorithms to switch back and forth eliminates certain bit in iterative process by the phenomenon that repeatedly overturns to improve decoding success rate, wherein a kind of algorithm is as main algorithm, another algorithm is as auxiliary algorithm, thus when complexity is substantially constant, this algorithm more existing other weighted bits upset algorithm improves the coding gain of 0.8 ~ 1.2dB, be applicable to long and short code, Regular codes and Irregular codes.

Object of the present invention is realized by following technical scheme:

Based on the LDPC interpretation method of the mixed weighting bit reversal that loop is eliminated, choose two existing weighted bit upset algorithms to mix, described main algorithm adopts existing weighted bit to overturn performance preferably algorithm in algorithm, and its effect is the repeatedly wrong flop phenomenon that the Initial Decoding after carrying out code word hard decision produces because of loop with the auxiliary algorithm of elimination.The secondary good algorithm of described auxiliary algorithm picks performance, its effect is the repeatedly wrong flop phenomenon that the main algorithm of elimination produces because of loop.In decode procedure, first decoder carries out hard decision to the code word received, and obtains the input information (as shown in Figure 1) of the codeword sequence after adjudicating as decoding.Arrange maximum iteration time, it act as the maximum arranging in decode procedure and allow iterations simultaneously, and finally whether success can terminate to make decoding.Main algorithm described herein adopts RR-WBF algorithm, and described auxiliary algorithm adopts IM-WBF algorithm; The expression formula of described RR-WBF is:

$E_n^1=\frac{1}{\left|y_n\right|}\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)\left(\underset{n^{\′}\∈N\left(m\right)}{\Σ}\right|y_{n^{\′}}\left|\right)$

The expression formula of described IM-WBF algorithm is:

$E_n^2=\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)w_{n,m}-\mathrm{\α}\·\left|y_n\right|;$

Wherein, | y _n| represent the absolute value of channel output valve.S _mrepresent the syndrome of check matrix.α is numerical value and adjustable, $\begin{array}{cc}{w}_{n,m}=\underset{i\∈N\left(m\right)\backslash n}{\min }\left|y_i\right|& n\∈N\left(m\right)\end{array},m=1,2,...,W.$

Interpretation method of the present invention comprises the following steps:

Step 1: decoder initialization, comprises H matrix and reads in, and receives the hard decision of code word and calculates for the relevant parameter of step 3, step 5 and step 7, namely for calculating the judgement standard En of main algorithm and auxiliary algorithm ¹, En ²relevant parameter.

Step 2: the codeword sequence verifying current input, the i.e. syndrome of calculation check matrix, if syndrome is full 0, successfully decoded and terminate; Otherwise, enter step 3.

Step 3: according to the relevant parameter of step 1 and the syndrome result of step 2, calculate main algorithmic decision standard En ¹and overturn the bit z of most probable mistake ₁, enter step 4 simultaneously.

Step 4: constantly detect in iterative process and judge the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical.If identical, enter step 5; If not identical, proceed iteration.

Step 5: again to the syndrome result of codeword sequence after hard decision according to the relevant parameter of step 1 and step 2, calculate auxiliary algorithmic decision standard En ²and and overturn the bit z of most probable mistake ₂, enter step 6 simultaneously.

Step 6: constantly detect in iterative process and judge the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical.If identical, enter step 7; If not identical, proceed iteration.

Step 7: continue to step 5 process after codeword sequence calculate main algorithmic decision standard E _n ¹and overturn the bit z of most probable mistake ₁.

Repeat step 2 to step 7, when successfully decoded or stop iteration when reaching maximum iteration time, and export coding sequence or export decoding failure information.

It is emphasized that: step 5 is again adjudicated and bit reversal with auxiliary algorithm starting most the codeword sequence after hard decision.Main cause is now for main algorithm, and the ring in receiving sequence check matrix has made decoding occur endless loop, i.e. the upset of mistake repeatedly of certain bit.And now receiving sequence is for auxiliary algorithm, the ring of check matrix also fails to make decoding occur endless loop, so utilize auxiliary algorithm to carry out decoding again to receiving sequence from, can be very fast successfully decoded.If but produce the phenomenon of certain bit mistake upset repeatedly at main algorithm after directly to main algorithm process after the auxiliary algorithm of codeword sequence carry out decoding, some loop will be caused to destroy, thus cannot decoding success.And step 7 is that the codeword sequence after step 5 processes directly is re-used to main algorithm and carries out decoding, although as some loop above-mentioned cannot destroy, but this step is compromised on performance, complexity and decoding speed, ensure that decoding performance and the decoding speed of LB-MWBF.

The bit z that current iteration overturns is detected in described iterative process _kthe bit z overturn with last iteration _k-1time identical, the loop described in now check matrix has made decoding occur the phenomenon of certain bit mistake upset repeatedly.

Described check matrix H has the capable L row of W, H _mnrepresent the n-th element that check matrix m is capable.

The method calculating check matrix syndrome in described step 2 is:

$s_m={\mathrm{\Σ}}_{n=1}^L{z}_n\·H_{mn},m=1,2,...,W$

In described step 3 and step 7, the bit z of described location most probable mistake ₁method be:

$z_1=\arg {\max }_{1\≤n\≤L}{E}_n^1$

In described step 5, the bit z of described location most probable mistake ₂method be:

$z_2=\arg {\max }_{1\≤n\≤L}{E}_n^2$

Wherein, E _n ¹be the judgement standard value of main algorithm, E _n ²it is the judgement standard value of auxiliary algorithm.

Operation principle of the present invention: the reason of most of weighted bit upset algorithm decoding failure is that the loop in check matrix makes decoding be absorbed in error bit and repeatedly judges failed circulation.What obtain after the code word obtained after being embodied in kth time iteration and before certain iteration is identical, then the secondary later computing of kth has all been absorbed in an endless loop.And the bit z showing as current iteration upset the most general _kthe bit z overturn with last iteration _k-1identical.For a specific receiving sequence, because path of information flow in different algorithm decode procedures is different, the loop in check matrix is also just different on the impact of decoding effect.The method adopting decode procedure two kinds of algorithms to switch back and forth eliminates the upset of mistake repeatedly that bit reversal in decode procedure produces, and wherein a kind of algorithm is as main algorithm, and another algorithm is as auxiliary algorithm.Such as, when main algorithm is absorbed in mistake upset repeatedly, re-starts decoding by auxiliary algorithm, main algorithm can be helped to break the circulation of mistake; Otherwise, when auxiliary algorithm is absorbed in mistake upset repeatedly, carry out continuation decoding by switching main algorithm, the wrong circulation that again helped auxiliary algorithm to break.And utilize auxiliary algorithm from, to carry out this important method of decoding again to receiving sequence, the loop produced in main algorithm can be broken away from more thoroughly, thus can be very fast successfully decoded.

Compared with prior art, tool has the following advantages and beneficial effect in the present invention:

(1) when overall complexity changes little, more existing other weighted bits upset algorithm improves the coding gain of 0.8 ~ 1.2dB.

(2) the present invention is while improve coding gain, and mean iterative number of time has had certain minimizing, and convergence rate is very fast, and implementation is relatively simple.

(3) the present invention is using weighted bit upset algorithm as core algorithm, and compare with soft-decision decoding method, computation complexity is low, and decoding performance is close to soft-decision decoding method further.

Accompanying drawing explanation

Fig. 1 is the workflow diagram of the LDPC interpretation method of the mixed weighting bit reversal based on loop elimination of the present invention.

Fig. 2 is one of foundation of the present invention: the formula construction unit comparison diagram of several existing weighted bit upset algorithmic decision standard.

Fig. 3 is foundation two of the present invention: several existing weighted bit upset algorithm is to (L, d _v, d _c) the complexity comparison diagram of each iteration of canonical LDPC code decoding.

Fig. 4 is simulating, verifying figure of the present invention: the Block Error Rate contrast schematic diagram of (500,250) canonical LDPC code.

Fig. 5 is simulating, verifying figure of the present invention: the mean iterative number of time contrast schematic diagram of (500,250) canonical LDPC code.

Fig. 6 is simulating, verifying figure of the present invention: the Block Error Rate contrast schematic diagram of (1008,504) irregular LDPC codes.

Fig. 7 is simulating, verifying figure of the present invention: the mean iterative number of time contrast schematic diagram of (1008,504) irregular LDPC codes.

Fig. 8 is simulating, verifying figure of the present invention: the Block Error Rate contrast schematic diagram of (2048,1018) irregular LDPC codes.

Fig. 9 is simulating, verifying figure of the present invention: the mean iterative number of time contrast schematic diagram of (2048,1018) irregular LDPC codes.

Figure 10 is simulating, verifying figure of the present invention: the Block Error Rate contrast schematic diagram of (4000,2000) canonical LDPC code.

Figure 11 is simulating, verifying figure of the present invention: the mean iterative number of time contrast schematic diagram of (4000,2000) canonical LDPC code.

Embodiment

Below in conjunction with embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not limited thereto.

Based on the LDPC interpretation method (Loop-BreakbasedMixedWeightedBit-Flipping of the mixed weighting bit reversal that loop is eliminated, LB-MWBF), that algorithm of the present invention is to (L as shown in Figure 1, K) embodiment of LDPC code, wherein L represents code word size, and K represents information bit length.Setting H=[H _mn] be the check matrix of LDPC.N (m)={ n:H _mn=1} represents all bit nodes participating in check-node m, M (n)={ m:H _mn=1} indicates all check-nodes that bit node n participates in.N (m) n represent the every other bit node participating in check-node m except n, M (n) m represent the every other check-node that except m bit node n participates in.D is collectively expressed as the quantity of " 1 " in canonical LDPC check matrix often row _c, often in row, the quantity of " 1 " is collectively expressed as d _v.

If LDPC code word c={c ₁, c ₂..., c _lapply in white Gaussian noise (AWGN) channel, modulation system is BPSK, is mapped as transfer sequence x={x ₁, x ₂..., x _l, wherein x _n=2c _n-1, the codeword sequence received is x+w=y={y ₁, y ₂..., y _l, wherein y _n=x _n+ w _n, w _nfor independently white Gaussian noise variable.Bit sequence after decoding is z={z ₁, z ₂..., z _l.

The present embodiment chooses RR-WBF as main algorithm, and IM-WBF is described as the embodiment of auxiliary algorithm to LB-MWBF:

Step 1: as shown in the step 1 in Fig. 1, decoder initialization, comprises H matrix and reads in, and receives the hard decision of code word and calculates the judgement standard En being used for main algorithm and auxiliary algorithm ¹, En ²relevant parameter.According to the judgement standard formula of algorithm, need to calculate:

$\begin{array}{cc}{w}_{n,m}=\underset{i\∈N\left(m\right)\backslash n}{\min }\left|y_i\right|& n\∈N\left(m\right)\end{array},m=1,2,...,W$

Step 2: as shown in step 2 in Fig. 1, verifies the codeword sequence of current input, i.e. the syndrome of calculation check matrix if syndrome is full 0, successfully decoded and terminate; Otherwise, enter step 3.

Step 3: as shown in step 3 in Fig. 1, calculates main algorithmic decision standard En ¹and overturn the bit z of most probable mistake ₁, enter step 4 simultaneously.

Wherein, main algorithm RR-WBF judgement standard $E_n^1=\frac{1}{\left|y_n\right|}\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)\left(\underset{n^{\′}\∈N\left(m\right)}{\Σ}\right|y_{n^{\′}}\left|\right);$

The method of the bit of location most probable mistake is:

Step 4: as shown in step 4 in Fig. 1, constantly detects in iterative process and judges the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical.If identical, enter step 5; If not identical, proceed iteration.

Step 5: as shown in the step 5 in Fig. 1, calculates auxiliary algorithmic decision standard En to codeword sequence after hard decision again ²and and overturn the bit z of most probable mistake ₂, enter step 6 simultaneously.

Wherein, auxiliary algorithm IM-WBF judgement standard $E_n^2=\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)w_{n,m}-\mathrm{\α}\·\left|y_n\right|;$

The method of the bit of location most probable mistake is:

Step 6: as shown in the step 6 in Fig. 1, constantly detects in iterative process and judges the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical.If identical, enter step 7; If not identical, proceed iteration.

Step 7: as shown in the step 7 in Fig. 1, continue to step 5 process after codeword sequence calculate main algorithmic decision standard E _n ¹and overturn the bit z of most probable mistake ₁.

Repeat step 2 to step 7, when successfully decoded or stop iteration when reaching maximum iteration time, and export coding sequence or export decoding failure information.

The reason that most of weighted bit overturns algorithm decoding failure is that the loop in check matrix makes decoding be absorbed in error bit and repeatedly judges failed circulation.What obtain after the code word obtained after being embodied in kth time iteration and before certain iteration is identical, then the secondary later computing of kth has all been absorbed in an endless loop.And the bit z showing as current iteration upset the most general _kthe bit z overturn with last iteration _k-1identical.Due to the criterion difference (as shown in Figure 2) of different interpretation method, for a specific receiving sequence, the loop in check matrix is also just different on the impact of decoding effect.And the method that LB-MWBF algorithm adopts RR-WBF algorithm and IM-WBF algorithm to switch back and forth eliminates the upset of mistake repeatedly that bit reversal in decode procedure produces.Such as, when main algorithm is absorbed in mistake upset repeatedly, re-starts decoding by auxiliary algorithm, main algorithm can be helped to break the circulation of mistake; Otherwise, when auxiliary algorithm is absorbed in mistake upset repeatedly, carries out continuation decoding by switching main algorithm, again helping auxiliary algorithm to break the circulation of mistake.And utilize auxiliary algorithm from, to carry out this important method of decoding again to receiving sequence, the loop produced in main algorithm can be broken away from more thoroughly, thus can be very fast successfully decoded.LB-MWBF algorithm improves coding gain, accelerates decoding speed, compares algorithm complex substantially constant (as shown in Figure 3) with RR-WBF with IM-WBF algorithm simultaneously.

When code word application in white Gaussian noise (AWGN), when modulation system is BPSK, with (500,250) Regular codes, (1008,504) Irregular codes, (2048,1018) Irregular codes, (4000,2000) Regular codes are example, the performance of algorithm such as upset such as weighted bit such as contrast LB-MWBF and LC-WBF, IM-WBF, RR-WBF etc. and mean iterative number of time.

(1) as shown in Figure 4 and Figure 5, for (500,250) Regular codes, in the error rate 10 ^-4near, LB-MWBF compares RR-WBF 0.8dB gain, has the gain higher than 1.2dB compared to IM-WBF and LC-WBF.And the mean iterative number of time of LB-MWBF also has certain reduction compared to RR-WBF, be only about 3/4 of IM-WBF and LC-WBF, namely iterative convergence speed has had certain increase simultaneously.

(2) as shown in Figure 6 and Figure 7, for (1008,504) Irregular codes, in the error rate 10 ^-4near, LB-MWBF compares RR-WBF 0.8dB gain, has the gain of 1.2dB compared to IM-WBF, larger compared to the gain of LC-WBF.And the mean iterative number of time of LB-MWBF also reduces about 10% compared to RR-WBF and IM-WBF, be only about 2/3 of LC-WBF, namely iterative convergence speed has had certain increase simultaneously.

(3) as shown in Figure 8 and Figure 9, for (2048,1018) Irregular codes, in the error rate 10 ^-4near, LB-MWBF compares RR-WBF and IM-WBF 0.8dB gain, larger compared to the gain of LC-WBF.And the mean iterative number of time of LB-MWBF also reduces about 10% compared to RR-WBF and IM-WBF, be only with LC-WBF about 2/3, namely iterative convergence speed has had certain increase simultaneously.

(4) as shown in Figure 10 and Figure 11, for (4000,2000) Regular codes, in the error rate 10 ^-4near, LB-MWBF compares RR-WBF and IM-WBF 0.8dB gain, larger compared to the gain of LC-WBF.And the mean iterative number of time of LB-MWBF also has certain reduction compared to RR-WBF, IM-WBF and LC-WBF, namely iterative convergence speed has had certain increase.

It is emphasized that, although all better compared to RR-WBF, IM-WBF and LC-WBF on the performance of LB-MWBF and iterative convergence speed, but it is identical that its computation complexity and other weighted bits overturn algorithm, simultaneously because LB-MWBF mean iterative number of time is lower than other weighted bits upset algorithm, so the computational load actually becomes in decode procedure will reduce further, iterative convergence speed is accelerated.

Above-described embodiment is the present invention's preferably execution mode; but embodiments of the present invention are not restricted to the described embodiments; change, the modification done under other any does not deviate from Spirit Essence of the present invention and principle, substitute, combine, simplify; all should be the substitute mode of equivalence, be included within protection scope of the present invention.

Claims (4)

1., based on the LDPC interpretation method of the mixed weighting bit reversal of loop elimination, it is characterized in that, comprise the following steps:

(1) decoder initialization: H matrix reads in, receives the hard decision of code word and the judgement standard En for calculating main algorithm and auxiliary algorithm ¹, En ²the calculating of relevant parameter;

(2) check matrix syndrome corresponding to LDPC code is calculated, if syndrome is full 0, successfully decoded and terminate; Otherwise, enter step (3);

(3) the syndrome result s that the relevant parameter calculated according to step (1) and step (2) calculate _m, calculate main algorithmic decision standard En ¹and overturn the bit z of most probable mistake ₁, enter step (4) simultaneously;

(4) in iterative process, constantly detection judges the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical: if identical, enter step (5); If not identical, proceed iteration;

(5) the syndrome result s that the relevant parameter calculated according to step (1) and step (2) calculate _m, again auxiliary algorithmic decision standard En is calculated to codeword sequence after hard decision ²and overturn the bit z of most probable mistake ₂, enter step (6) simultaneously;

(6) in iterative process, constantly detection judges the bit z that current iteration overturns _kthe bit z overturn with last iteration _k-1whether identical: if identical, enter step (7); If not identical, proceed iteration;

(7) continuation calculates main algorithmic decision standard E to the codeword sequence after step (5) process _n ¹and overturn the bit z of most probable mistake ₁;

(8) repeat step (2) to (7), when successfully decoded or stop iteration when reaching maximum iteration time, and export coding sequence or export decoding failure information.

2. the LDPC interpretation method of the mixed weighting bit reversal based on loop elimination according to claim 1, is characterized in that; In step (2), the check matrix syndrome that described calculating LDPC code is corresponding, is calculated by following formula:

$s_m={\mathrm{\Σ}}_{n=1}^L{z}_n\·H_{mn},m=1,2,...,W;$

Wherein check matrix has the capable L row of W, H _mnrepresent the n-th element that check matrix m is capable, z _nthe bit value that after expression decoding, position n is corresponding.

3. the LDPC interpretation method of the mixed weighting bit reversal based on loop elimination according to claim 1, it is characterized in that: described main algorithm adopts RR-WBF algorithm, described auxiliary algorithm adopts IM-WBF algorithm;

The expression formula of described RR-WBF algorithm is:

$E_n^1=\frac{1}{\left|y_n\right|}\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)\left(\underset{n^{,}\∈N\left(m\right)}{\Σ}\right|y_n\left|\right),$

Wherein, N (m)={ n:H _mn=1}, represents the set of all bit node compositions of participation m check-node; M (n)={ m:H _mn=1}, represents the set of all check-node compositions that the n-th bit node participates in, s _mrepresent the syndrome that m check-node is corresponding, | y _n| represent the absolute value of the n-th bit node channel output valve;

The expression formula of described IM-WBF algorithm is:

$E_n^2=\underset{m\∈M\left(n\right)}{\Σ}(2s_m-1)w_{n,m}-\mathrm{\α}\·\left|y_n\right|,$

Wherein, M (n)={ m:H _mn=1}, represents the set of all check-node compositions that the n-th bit node participates in, s _mrepresent the syndrome that m check-node is corresponding, | y _n| represent the absolute value of the n-th bit node channel output valve, α is numerical value;

$\begin{array}{ccc}{w}_{n,m}=\underset{i\∈N\left(m\right)\backslash n}{min}\left|y_i\right|& n\∈N\left(m\right)& m=1,2,...,W\end{array},$

Wherein, if N (m) set comprises the n-th bit node, then N (m) n represent the n-th bit node is deleted from N (m) set after the set of gained, | y _i| represent the absolute value of i-th bit node channel output valve, w _n,mfor N (m) the minimum value of absolute value of bit node channel output valve in n set.

4. the LDPC interpretation method of mixed weighting bit reversal eliminated based on loop according to claim 1, is characterized in that, in step (3) and step (7), and the bit of described most probable mistake, its localization method is:

$z_1=\arg {\max }_{1\≤n\≤L}{E}_n^1$

represent from the bit position of 1 to L, choose main algorithmic decision standard value E _n ¹the bit position that maximum is corresponding;

In step (5), the bit of described most probable mistake, its localization method is:

$z_2=\arg {\max }_{1\≤n\≤L}{E}_n^2$

represent from the bit position of 1 to L, choose the judgement standard value E of auxiliary algorithm _n ²the bit position that maximum is corresponding.