CN101188427B - Confidence spreading decoding method for enhancing error correction - Google Patents

Confidence spreading decoding method for enhancing error correction Download PDF

Info

Publication number
CN101188427B
CN101188427B CN2007101355909A CN200710135590A CN101188427B CN 101188427 B CN101188427 B CN 101188427B CN 2007101355909 A CN2007101355909 A CN 2007101355909A CN 200710135590 A CN200710135590 A CN 200710135590A CN 101188427 B CN101188427 B CN 101188427B
Authority
CN
China
Prior art keywords
decoding
variable node
node
degree
confidence
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN2007101355909A
Other languages
Chinese (zh)
Other versions
CN101188427A (en
Inventor
高西奇
周沐
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huawei Technologies Co Ltd
Original Assignee
Southeast University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Southeast University filed Critical Southeast University
Priority to CN2007101355909A priority Critical patent/CN101188427B/en
Publication of CN101188427A publication Critical patent/CN101188427A/en
Application granted granted Critical
Publication of CN101188427B publication Critical patent/CN101188427B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Error Detection And Correction (AREA)

Abstract

The invention provides an encoding method for a low-density parity-check code in a wireless communication system. The encoding method relates to a confidence coefficient spreading decoding with enhanced cascade correction after the initial confidence coefficient spreading decoding. After initial confidence coefficient spreading decoding, the invention selects out the variable nodes needing to be corrected, assigns a value more than the present value to the variable nodes compulsorily, tests all the combined branches of the variable nodes, conducts confidence coefficient spreading decoding respectively on each combined branch, and then again selects out variable nodes needing to be corrected on each branch. So repeatedly, the selected variable nodes and branches form an encoding tree; different combinations of variable nodes selecting method and encoding tree can form different encoders; the complexity and the performance of the encoder can be controlled by adjusting the layer number of the tree and the spreading encoding cycle times of the confidence on the branches.

Description

The confidence degree propagation interpretation method that error correction strengthens
Technical field
The present invention relates to be used for the interpretation method of wireless communication system loe-density parity-check code, the confidence degree propagation interpretation method that especially a kind of error correction strengthens.
Background technology
Confidence degree propagation interpretation method is applied to the decoding of loe-density parity-check code (LDPC Codes) and turbine code (Turbo Codes).The process of decoding is to do the message transmission in Tanner figure.If do not encircle among the figure, confidence degree propagation interpretation method is a kind of MAP decoding method.If encircle and exist among the figure, confidence degree propagation interpretation method is a kind of near MAP decoding method.There are a lot of short circles in the drawings in loe-density parity-check code with good minimum distance characteristic.A code length is less relatively than the possibility that the short circle of long loe-density parity-check code exists.And code length makes it be worse than the performance of maximum-likelihood decoding far away than a large amount of short performances that have a strong impact on degree of confidence propagation decoding of enclosing in the short loe-density parity-check code.
Adopt reliability statistics ordering interpretation method can effectively reduce the degree of confidence propagation decoding of short code and the performance gap between the maximum-likelihood decoding in conjunction with confidence degree propagation interpretation method.But because need the gaussian elimination and the ordering of high complexity, reliability statistics ordering decoding is difficult to practical application.
Summary of the invention
The invention provides the confidence degree propagation interpretation method that a kind of error correction strengthens, can replace reliability statistics ordering interpretation method.Because do not have gaussian elimination and ordering (when only selecting a variable node) at every turn, the complexity of the inventive method significantly reduces than reliability statistics ordering interpretation method.The inventive method can reach the same decoding performance with reliability statistics ordering interpretation method, and can be by the number of plies of adjusting tree and the complexity and the performance of the propagation decoding cycle-index control decoder of putting letter in the branch.
The technical scheme of the inventive method is: the degree of confidence propagation decoding that the cascade error correction strengthens behind initial degree of confidence propagation decoding, it is characterized in that earlier through initial degree of confidence propagation decoding, select the variable node (one or more) that needs error correction, pressure gives the value of these variable nodes greater than currency (positive or negative sufficiently stable value), test these variable nodes positive negative value all may make up branch, in each combination branch, carry out degree of confidence propagation decoding respectively, next in each branch, select the variable node (one or more) that needs error correction again, so continue, variable node that these are chosen and branch have constituted a decoding tree; Adopt the combination of different variable node systems of selection and decoding tree, constitute different decoders; By the number of plies of adjusting tree and the complexity and the performance of the propagation decoding cycle-index control decoder of putting letter in the branch.
The system of selection of variable node has following three kinds:
Method 1, step is as follows:
A) in the degree of confidence propagation decoding process, add up the upset number of times of non-selected variable node;
B) after degree of confidence propagation decoding is finished, according to the upset number of times from high to low, choice variable node in non-selected variable node set.
Method 2, step is as follows:
A) after degree of confidence propagation decoding is finished, in non-selected variable node, find out the check-node and all variable nodes that are connected with these check-nodes that do not satisfy constraint;
B) add up the not check-node number (the mistake number of degrees) of satisfied constraint that each variable node of finding out connects, and find out the maximum wrong number of degrees.In non-selected variable node set, find out the variable node that satisfies the wrong number of degrees=maximum wrong number of degrees;
C) according to initial channel information absolute value order from small to large choice variable node from the variable node that step b) obtains.
Method 3, step is as follows:
A) in the degree of confidence propagation decoding process, add up the upset number of times of non-selected variable node;
B) after degree of confidence propagation decoding is finished, according to the upset number of times from high to low, in non-selected variable node set, find out one or more variable node.If maximum upset number of times 〉=2 enter step f);
C) in non-selected variable node, find out the check-node and all variable nodes that are connected with these check-nodes that does not satisfy constraint;
D) add up the check-node number (the mistake number of degrees) that does not satisfy constraint that each variable node of finding out connects, find out the maximum wrong number of degrees.The variable node of the number of degrees=maximum wrong number of degrees locates errors in the set of non-selected variable node;
E) from the variable node that step d) obtains, find out variable node according to initial channel information absolute value order from small to large;
F) select the variable node found out.
The decode procedure of decoding tree, step is as follows:
Step 1: initialization, carry out initial degree of confidence propagation decoding;
Step 2: on the current tree node that arrives, carry out variable node and select, following several possibility situation is arranged:
If a) all branch's decodings finish, enter step 5;
B) if arrive leaf node (the present tree number of plies j 〉=maximal tree number of plies J in this path Max), father node is returned, present tree number of plies j=j-1 in the decoding path.If father node has kept side information, recovering side information is the side information that father node keeps, and other branches of test father node, enters step 3;
C) except that top situation, use variable node system of selection 1 or method 2 or method 3 choice variable nodes, force to give the value of these variable nodes greater than currency.
Step 3: test all combination branches of the positive negative value of these variable nodes, execution in step 4 in each branch;
Step 4: in the branch of tree, carry out degree of confidence propagation decoding, two kinds of selections are arranged:
A) the initialization side information carries out degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2;
B) keeping previous step (present node place) side information, proceed degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2.
Step 5: following may the selection arranged:
A) adopt greedy method.Degree of confidence propagation decoding success, with the estimation code word that translates as final result.Decoding finishes;
B) adopt tabulating method.If also have the path not finish decoding, return step 2.Otherwise in the estimation code word of all confidence spread decoding success, select one as final result.Decoding finishes.
Above-mentioned each variable node system of selection and a decoding tree combination can constitute a kind of decoder, can constitute three kinds of decoders altogether.Also can in three kinds of decoders, choose wantonly its two or select all as sub-decoder, connect or the new decoder of formation connected in series by parallel, decode results in the result of sub-decoder according to the priority criterion or translate the time criterion the soonest and obtain.
Advantage of the present invention and beneficial effect:
1. compare with reliability statistics ordering interpretation method, the inventive method decoding complexity reduces greatly, and decoding performance is but outstanding equally.
2. compare with reliability statistics ordering interpretation method, the complexity control of the inventive method is more flexible.The inventive method can be by regulating tree the number of plies and complexity and the performance of controlling decoder arbitrarily of the propagation decoding cycle-index of putting letter in the branch.
3. compare with reliability statistics ordering interpretation method, the inventive method can be exported soft information, is applicable to the associating iterative detection.
4. compare with existing confidence level interpretation method, method of the present invention can significantly promote from short code to the degree of confidence propagation decoding performance of long code.The wrong flat bed that occurs in the time of can effectively removing decoding with low complexity.
5. the inventive method can constitute new decoder device by the parallel or serial combination of sub-decoder, reaches better decoding performance.
The inventive method can with the combination of improving one's methods of any degree of confidence propagation decoding.
The present invention is applicable to various loe-density parity-check codes and can uses the turbine code of degree of confidence propagation decoding, mainly comprises:
1.MacKay loe-density parity-check code.
2.Gallager loe-density parity-check code.
3.Progressive Edge Growth loe-density parity-check code.
4.Quasi-Cyclic loe-density parity-check code.
The loe-density parity-check code of (5.WiMAX IEEE Std 802.16e) agreement regulation.
6.DVB-S2 the regulation Quasi-Cyclic sign indicating number of standard.
7. the existing turbine code that can use confidence degree propagation interpretation method.
8.Finite Geometry loe-density parity-check code.
Description of drawings
Fig. 1 is the decoding architecture figure of decoding tree;
Fig. 2 is the combinational decoder scheme of the inventive method:
(a) be the parallel decoder that connects;
(b) be decoder connected in series.
Embodiment
Below in conjunction with accompanying drawing the inventive method is described in detail:
Fig. 1 is one 2 fork decoding tree.E represents the side information of the degree of confidence propagation decoding at tree node place.v SelThe variable node of representing the tree node place to select.J represents the number of plies of decoding tree.k jRepresent the maximum cycle of the degree of confidence propagation decoding in the j layer tree branch.
With the loe-density parity-check code is example.The sparse check matrix of definition loe-density parity-check code is H=[h Ij] MxnThis matrix can represent that wherein matrix column is represented with variable node with a two-dimensional plot, and row is represented with check-node.We represent the variable node set with V, and C represents the check-node set.Be connected to check-node c jVariable node set be
V(c j)＝{v i|h ij＝1，0≤i≤n-1}。
Be connected to variable node v iCheck-node set be
C(v i)＝{c j|h ij＝1，0≤j≤m-1}。
The code word x=[x that has encoded 0, x 1..., x N-1] expression.Corresponding receiving sequence y=[y 0, y 1..., y N-1] expression.The verification vector is defined as
b＝[b 0，b 1，...，b m-1]＝xH T
If b j≠ 0, check-node c jDo not satisfy constraint.Otherwise, check-node c jSatisfy constraint.The initial channel information Q of the variable node of degree of confidence propagation decoding i 0, 0≤i≤n-1 represents.In the k time circulation of degree of confidence propagation decoding, variable node v iLog-likelihood ratio Q i kExpression.From check-node c j, c j∈ C (v i) to variable node v iThe limit message table be shown r Ji kFrom variable node v i, v i∈ V (c j) to check-node c jThe limit message table be shown q Ij kHard decision is defined as
x ^ i k = 1 , if Q i k > 0 0 , else , 0 ≤ i ≤ n - 1
If b k = x ^ k H T = 0 Perhaps k 〉=k Max, degree of confidence propagation decoding stops.Estimate code word
Figure S2007101355909D00043
Export as final code word.Otherwise decoding continues.
Degree of confidence propagation decoding is after k circulation, and the check-node set (SUC) of not satisfying constraint is expressed as C S k ⊆ C . Insecure variable node is defined as and C S kIn the check-node variable node that in the Tanner of sign indicating number figure, has at least a limit to link to each other.Insecure variable node set is expressed as V S k ⊆ V . SUC schemes G SBe by SUCC S kThe figure that induces.Variable node v ∈ V S k At G SIn the wrong number of degrees (the check-node number that does not satisfy constraint of connection) be expressed as d Gs k(v).Maximum wrong number of degrees d Gs k(v) be expressed as d Gs K, maxSatisfy the wrong number of degrees=maximum wrong number of degrees, promptly d G S k ( v ) = d G S k , max Variable node set be expressed as S v K, max
Variable node v iThe upset number of times be defined as
F v , i 0 = 0 , F v , i k = F v , i k - 1 , ifsign ( Q i k ) = sign ( Q i k - 1 ) F v , i k - 1 + 1 otherwise k ≥ 1
Maximum upset number of times is defined as
F v k , max = max v i ∈ V F v , i k
In the degree of confidence propagation decoding that error correction strengthens, selecteed variable node set is expressed as V Sel, non-selected variable node set is expressed as V Unsel
The system of selection of variable node:
Method 1, step is as follows:
A) carry out seven degree of confidence propagation decoding circulations, add up each variable node V i∈ V Unsel, upset number of times F V, i k
B) after k degree of confidence propagation decoding finished, according to upset number of times F V, i kFrom high to low, from variable node v i∈ V UnselIn select one or more variable node successively.
Method 2, step is as follows:
A) after k degree of confidence propagation decoding finished, find out SUC figure and and SUCC S kThe variable node set V that connects S k
B) statistical variable node v i ∈ V S k ∩ V unsel D Gs k(v i).Find out the variable node set of satisfying the wrong number of degrees=maximum wrong number of degrees
S v k , max = { v i ∈ V S k ∩ V unsel : d G S k ( v ) = d G S k , max } ⊆ V S k ∩ V unsel ;
C) according to initial channel information absolute value | Q i 0| order from small to large is from variable node v i ∈ S v k , max ∩ V unsel In select one or more variable node successively.
Method 3, step is as follows:
A) carry out k degree of confidence propagation decoding circulation, add up each variable node v i∈ V UnselUpset number of times F V, i k
B) after seven degree of confidence propagation decoding are finished, according to upset number of times F V, i kFrom high to low, from variable node v i∈ V UnselIn select one or more variable node successively.If F v k , max ≥ 2 , Enter step f);
C) find out SUC figure and and SUCC S kThe variable node set V that connects S k
D) statistical variable node v i ∈ V S k ∩ V unsel D Gs k(v i).Find out the variable node set of satisfying the wrong number of degrees=maximum wrong number of degrees
S v k , max = { v i ∈ V S k ∩ V unsel : d G S k ( v ) = d G S k , max } ⊆ V S k ∩ V unsel ;
E) according to initial channel information absolute value | Q i 0| order from small to large is from variable node v i ∈ S v k , max ∩ V unsel In find out one or more variable node successively;
F) select one or more variable node found out.
The decoding of decoding tree:
The decode procedure of decoding tree is to be handled by the step of two on many tree nodes to constitute:
A) use one of above 3 kinds of variable node systems of selection to select one or more variable nodes that need error correction;
B) force to give these variable nodes positive or negative sufficiently stable value ± S.Test the branch that to make up of the positive negative value of these variable nodes, in each combination branch, carry out degree of confidence propagation decoding respectively.
If only select a variable node in the variable node system of selection at every turn, decoding tree just constitutes one 2 fork decoding tree.If select t variable node at every turn, decoding tree constitutes one 2 tThe fork decoding tree.In ensuing explanation, can see 2 tThe fork decoding tree can be represented with 2 fork decoding trees.
We are example with 2 fork decoding trees.In the variable node system of selection, only select a variable node at every turn.Here decoding path definition is the paths from the root node of decoding tree to leaf node.
Decoding tree decoding serial decoding scheme
The decoding step is as follows:
Step 1: initialization.Present tree number of plies j=0.Carry out initial degree of confidence propagation decoding;
Step 2: present tree number of plies j=j+1.On the current tree node that arrives, carry out variable node and select, following several possibility situation is arranged:
If a) all path decodings finish or the degree of confidence propagation decoding success, enter step 5;
B) if all sub-branches of current tree node are tested, the variable node cancellation of having selected on the current tree node is selected, and returns the father node of current tree node, present tree number of plies j=j-1;
C) if arrive leaf node (the present tree number of plies j 〉=maximal tree number of plies J in this path Max), father node is returned, present tree number of plies j=j-1 in the decoding path.If father node has kept side information, recovering side information is the side information that father node keeps, and other branches of test father node, enters step 3;
D) except that top situation, use variable node system of selection 1 or method 2 or method 3 to select a variable node, force to give these variable nodes positive or negative sufficiently stable value.
Step 3: test execution in step 4 in not tested in all branches that may constitute of positive negative value of these variable nodes the branch;
Step 4: in the branch of tree, carry out degree of confidence propagation decoding, two kinds of selections are arranged:
A) the initialization side information carries out degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2;
B) keeping previous step (present node place) side information, proceed degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2.
Step 5: following may the selection arranged:
A) adopt greedy method.Degree of confidence propagation decoding success, with the estimation code word that translates as final result.Decoding finishes;
B) adopt tabulating method.If also have the path not finish decoding, return step 2.Otherwise in the estimation code word of all confidence spread decoding success, select most possible one as final result.Decoding finishes.
In Fig. 1, the ergodic process of tree node is 1 → 2 → 3 → 4 → 5 → 6 → 7.k jIt is the maximum cycle of degree of confidence propagation decoding in the J layer branch of decoding tree.When we need select t＞1 variable node in the variable node system of selection, only need as 2 the pitching in the decoding trees of Fig. 1, make k j=0, t * l+1≤j≤t * (l+1)-1, l 〉=0.
Decoding tree decoding parallel decoding scheme
Begin along all possible paths from the root node of decoding tree, according in layer branch and tree node alternate treatment are simultaneously to the leaf node propelling, step is as follows:
Step 1: initialization.Present tree number of plies j=0.Carry out initial degree of confidence propagation decoding;
Step 2: present tree number of plies j=j+1.Select at the variable node that on each tree node of anterior layer, carries out that arrives, following several possibility situation arranged:
If a) present tree number of plies j 〉=maximal tree number of plies J Max, enter step 5;
B) if the degree of confidence propagation decoding success enters step 5;
C) except that top situation, use variable node system of selection 1 or method 2 or method 3 to select a variable node, pressure gives these variable nodes positive or negative sufficiently stable value, test all branches that may constitute of positive negative value of these variable nodes, execution in step 3 in each branch.
Step 3: on the branch of tree, carry out degree of confidence propagation decoding, two kinds of selections are arranged:
A) the initialization side information carries out degree of confidence propagation decoding, selects statistics for variable node simultaneously;
B) keep previous step (father node place) side information, proceed degree of confidence propagation decoding, select statistics for variable node simultaneously.
Step 4: finish if work as the branch process of all tree nodes of anterior layer, enter step 2.Otherwise execution in step 3 in not tested tree branch;
Step 5: following may the selection arranged:
1) adopts greedy method.Degree of confidence propagation decoding success, with the estimation code word that translates as final result.Decoding finishes;
2) adopt tabulating method.If also have the path not finish decoding, return step 2.Otherwise in the estimation code word of all confidence spread decoding success, select most possible one as final result.Decoding finishes.
In Fig. 1, the ergodic process of tree node is 1 → 2,5 → 3,4,6,7.Same serial scheme, when we need select t＞1 variable node in the variable node system of selection, only need as 2 the pitching in the decoding trees of Fig. 1, make k j=0, t * l+1≤j≤t * (l+1)-1, l 〉=0.
The combinational decoder scheme:
The degree of confidence propagation decoding that three kinds of error correction strengthen have a high regard for select its two or select all as sub-decoder, connect or the new decoder device of formation connected in series by parallel.Decode results is selected according to one of two kinds of criterions in the result of sub-decoder: priority criterion or translate the time criterion the soonest.
During parallel the connection with the parallel example that is connected to of three kinds of methods.The sequence of needs decoding is imported three sub-decoders simultaneously carry out the degree of confidence propagation decoding that error correction strengthens respectively.In son decoding, select decode results according to the priority criterion or according to translating the time criterion the soonest.The priority criterion: as (a) among Fig. 2, according to 1,2,3 priority obtains the result.If sub-decoder 1 has the result, the result who adopts sub-decoder 1 is as final result.Otherwise if sub-decoder 2 has the result, the result who adopts sub-decoder 2 is as final result.Otherwise if sub-decoder 3 has the result, the result who adopts sub-decoder 3 is as final result.According to translating the time criterion the soonest: as (a) among Fig. 2,1,2,3 sub-decoders are deciphered simultaneously, and which sub-decoder is successful decoding result's final result the most of just adopting which decoder earlier.
Be example with the connected in series of three kinds of methods when connected in series.As (b) among Fig. 2, the sequence of needs decoding is imported sub-decoder 1 earlier carry out the degree of confidence propagation decoding that error correction strengthens.If sub-decoder 1 has the result, the result who adopts sub-decoder 1 is as final result.Otherwise, the sequence input sub-decoder 2 of needs decoding is carried out the degree of confidence propagation decoding that error correction strengthens, if sub-decoder 2 has the result, the result who adopts sub-decoder 2 is as final result.Otherwise, the sequence input sub-decoder 3 of needs decoding is carried out the degree of confidence propagation decoding that error correction strengthens, if sub-decoder 3 has the result, the result who adopts sub-decoder 3 is as final result.
The present invention can be by the number of plies of adjusting tree and the complexity and the performance of the propagation decoding cycle-index control decoder of putting letter in the branch.Increase the number of plies of decoding tree and the confidence spread cycle-index in the branch and can improve decoding performance, but increased decoding complexity.Reduce the number of plies of decoding tree and the confidence spread cycle-index in the branch and can reduce complexity, but reduced decoding performance.Should the actual conditions when using determine the number of plies of decoding tree and the confidence spread cycle-index in the branch, the balance of acquisition decoding complexity and performance.

Claims (4)

1. the confidence degree propagation interpretation method that strengthens of error correction, it is characterized in that earlier through initial degree of confidence propagation decoding, select the variable node that needs error correction, pressure gives the value of these variable nodes greater than currency, test all combination branches of these variable nodes, in each combination branch, carry out degree of confidence propagation decoding respectively, next in each branch, select the variable node that needs error correction again, so continue, variable node that these are chosen and branch have constituted a decoding tree; Adopt the combination of different variable node systems of selection and decoding tree, constitute different decoders; By the number of plies of adjusting tree and the complexity and the performance of the propagation decoding cycle-index control decoder of putting letter in the branch.
2. the confidence degree propagation interpretation method that error correction according to claim 1 strengthens is characterized in that the system of selection of variable node has following three kinds:
Method 1, step is as follows:
A) in the degree of confidence propagation decoding process, add up the upset number of times of non-selected variable node;
B) after degree of confidence propagation decoding is finished, according to the upset number of times from high to low, choice variable node in non-selected variable node set;
Method 2, step is as follows:
A) after degree of confidence propagation decoding is finished, in non-selected variable node, find out the check-node and all variable nodes that are connected with these check-nodes that do not satisfy constraint;
B) adding up the check-node number that does not satisfy constraint that each variable node of finding out connects is the wrong number of degrees, and finds out the maximum wrong number of degrees, finds out the variable node that satisfies the wrong number of degrees=maximum wrong number of degrees in non-selected variable node set;
C) according to initial channel information absolute value order from small to large choice variable node from the variable node that step b) obtains;
Method 3, step is as follows:
A) in the degree of confidence propagation decoding process, add up the upset number of times of non-selected variable node;
B) after degree of confidence propagation decoding is finished, according to the upset number of times from high to low, in non-selected variable node set, find out one or more variable node, if maximum upset number of times 〉=2 enter step f);
C) in non-selected variable node, find out the check-node and all variable nodes that are connected with these check-nodes that does not satisfy constraint;
D) adding up the check-node number that does not satisfy constraint that each variable node of finding out connects is the wrong number of degrees, finds out the maximum wrong number of degrees, the variable node of the number of degrees=maximum wrong number of degrees that locate errors in non-selected variable node set;
E) from the variable node that step d) obtains, find out variable node according to initial channel information absolute value order from small to large;
F) select the variable node found out;
The decode procedure of decoding tree, step is as follows:
Step 1: initialization, carry out initial degree of confidence propagation decoding;
Step 2: on the current tree node that arrives, carry out variable node and select, following several possibility situation is arranged:
If a) all branch's decodings finish, enter step 5;
B) if arrive the leaf node in this path, present tree number of plies j 〉=maximal tree number of plies J Max, father node is returned, present tree number of plies j=j-1 in the decoding path; If father node has kept side information, recovering side information is the side information that father node keeps, and other branches of test father node, enters step 3;
C) except that top situation, use variable node system of selection 1 or method 2 or method 3 choice variable nodes, force to give the value of these variable nodes greater than currency;
Step 3: test all combination branches of the positive negative value of these variable nodes, execution in step 4 in each branch;
Step 4: in the branch of tree, carry out degree of confidence propagation decoding, two kinds of selections are arranged:
A) the initialization side information carries out degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2;
B) keeping previous step present node place side information, proceed degree of confidence propagation decoding, is variable node system of selection statistics simultaneously, enters step 2;
Step 5: following may the selection arranged:
A) adopt greedy method: the degree of confidence propagation decoding success, as final result, decoding finishes with the estimation code word that translates;
B) adopt tabulating method: if also have the path not finish decoding, return step 2, otherwise select one as final result in the estimation code word of all confidence spread decoding success, decoding finishes.
3. the confidence degree propagation interpretation method that error correction according to claim 2 strengthens is characterized in that each variable node system of selection and a decoding tree combination, constitutes a kind of decoder, constitutes three kinds of decoders altogether.
4. the confidence degree propagation interpretation method that error correction according to claim 3 strengthens, it is characterized in that in three kinds of decoders optional its two or select all as sub-decoder, connect or the new decoder of formation connected in series by parallel, decode results in the result of sub-decoder according to the priority criterion or translate the time criterion the soonest and obtain.
CN2007101355909A 2007-11-13 2007-11-13 Confidence spreading decoding method for enhancing error correction Active CN101188427B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2007101355909A CN101188427B (en) 2007-11-13 2007-11-13 Confidence spreading decoding method for enhancing error correction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2007101355909A CN101188427B (en) 2007-11-13 2007-11-13 Confidence spreading decoding method for enhancing error correction

Publications (2)

Publication Number Publication Date
CN101188427A CN101188427A (en) 2008-05-28
CN101188427B true CN101188427B (en) 2011-08-31

Family

ID=39480645

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2007101355909A Active CN101188427B (en) 2007-11-13 2007-11-13 Confidence spreading decoding method for enhancing error correction

Country Status (1)

Country Link
CN (1) CN101188427B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105099618A (en) * 2015-06-03 2015-11-25 香港中文大学深圳研究院 Decoding method based on physical network coding and corresponding data processing method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005065066A (en) * 2003-08-19 2005-03-10 Sony Corp Decoding device and method, and program
CN1859013A (en) * 2006-06-09 2006-11-08 东南大学 Low density odd-even check code iterative sequencing statistical decoding method
FR2894738A1 (en) * 2005-12-09 2007-06-15 Interface Sarl R Low density parity code decoding complexity reducing method for communication channel, involves decomposing low density parity check code decoding function into sub-functions represented by check and bit node units and memory controller
WO2007123257A1 (en) * 2006-04-21 2007-11-01 Kabushiki Kaisha Toshiba Method and apparatus for entropy coding and decoding

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005065066A (en) * 2003-08-19 2005-03-10 Sony Corp Decoding device and method, and program
FR2894738A1 (en) * 2005-12-09 2007-06-15 Interface Sarl R Low density parity code decoding complexity reducing method for communication channel, involves decomposing low density parity check code decoding function into sub-functions represented by check and bit node units and memory controller
WO2007123257A1 (en) * 2006-04-21 2007-11-01 Kabushiki Kaisha Toshiba Method and apparatus for entropy coding and decoding
CN1859013A (en) * 2006-06-09 2006-11-08 东南大学 Low density odd-even check code iterative sequencing statistical decoding method

Also Published As

Publication number Publication date
CN101188427A (en) 2008-05-28

Similar Documents

Publication Publication Date Title
CN1132320C (en) Optimal soft-output decoder for tail-biting trellis codes
TWI594583B (en) Gldpc soft decoding with hard decision inputs
AU2012321618B2 (en) Apparatus and method for transmitting and receiving data in communication/broadcasting system
CN110098838B (en) Error correction and erasure correction iterative decoding method of LDPC-RS product code
CN103269229B (en) A kind of mixed iteration interpretation method of LDPC-RS two dimension product code
US9831987B2 (en) Staggered parity
US20190273511A1 (en) Generation of spatially-coupled quasi-cyclic ldpc codes
CN101656541B (en) Coding method and device of RS codes
CN105763203B (en) Multi-element LDPC code decoding method based on hard reliability information
CN103929210B (en) Hard decision decoding method based on genetic algorithm and neural network
US20120060071A1 (en) Product code decoding method and device
US11082147B2 (en) Processing method, device and system for overlap multiplexing system
CN109981112A (en) A kind of sequencing statistical decoding method of partial cyclic redundancy check auxiliary
CN109379084B (en) Decoding method for burst errors
CN106452675B (en) A kind of spherical decoding method of polarization code
CN104052499A (en) Erasure correcting decoding method and system of LDPC code
CN101188427B (en) Confidence spreading decoding method for enhancing error correction
CN107947802A (en) The method and coder of rate-compatible encoding and decoding low-density parity check code
Ortega Espluga et al. New solutions to reduce the time-to-CED and to improve the CED robustness of the Galileo I/NAV message
CN104753542B (en) For the bit reversal of LDPC code and linear programming combination interpretation method
JP2009225325A (en) Decoding method and decoding apparatus, and program
US20210203364A1 (en) Apparatuses and methods for mapping frozen sets between polar codes and product codes
CN103731158B (en) A kind of building method of the LDPC check matrix for navigation signal
Espluga et al. New solutions to reduce the time-to-CED and to improve the CED robustness of the Galileo I/NAV message
CN105680882A (en) Hard decision decoding method for quadratic residue codes

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
TR01 Transfer of patent right

Effective date of registration: 20180925

Address after: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen

Patentee after: Huawei Technologies Co., Ltd.

Address before: 210096 No. four archway, 2, Jiangsu, Nanjing

Patentee before: Southeast University

TR01 Transfer of patent right