CN105978578B - The non-uniform quantizing method of low density parity check code and product decoding operation numerical value - Google Patents
The non-uniform quantizing method of low density parity check code and product decoding operation numerical value Download PDFInfo
- Publication number
- CN105978578B CN105978578B CN201610282279.6A CN201610282279A CN105978578B CN 105978578 B CN105978578 B CN 105978578B CN 201610282279 A CN201610282279 A CN 201610282279A CN 105978578 B CN105978578 B CN 105978578B
- Authority
- CN
- China
- Prior art keywords
- value
- uniform
- quantized
- quantizing
- quantization
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000013139 quantization Methods 0.000 claims abstract description 175
- 238000004088 simulation Methods 0.000 claims description 49
- 230000003247 decreasing effect Effects 0.000 claims description 5
- 241000208340 Araliaceae Species 0.000 claims description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 2
- 235000008434 ginseng Nutrition 0.000 claims description 2
- 238000012804 iterative process Methods 0.000 claims description 2
- 238000004891 communication Methods 0.000 abstract description 3
- 239000011159 matrix material Substances 0.000 description 8
- 230000007423 decrease Effects 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 5
- 238000011002 quantification Methods 0.000 description 4
- 238000012795 verification Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 208000019901 Anxiety disease Diseases 0.000 description 1
- 230000036506 anxiety Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M13/00—Coding, decoding or code conversion, for error detection or error correction; Coding theory basic assumptions; Coding bounds; Error probability evaluation methods; Channel models; Simulation or testing of codes
- H03M13/03—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words
- H03M13/05—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits
- H03M13/11—Error detection or forward error correction by redundancy in data representation, i.e. code words containing more digits than the source words using block codes, i.e. a predetermined number of check bits joined to a predetermined number of information bits using multiple parity bits
- H03M13/1102—Codes on graphs and decoding on graphs, e.g. low-density parity check [LDPC] codes
- H03M13/1148—Structural properties of the code parity-check or generator matrix
Landscapes
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The present invention relates to the non-uniform quantizing methods of low density parity check code and product decoding operation numerical value, belong to communication channel decoding technique field, this method are as follows: first to the Soft Inform ation of i-th of symbol of LDPC code, external information that check-node j is output to bit node i, the external information for being output to bit node i to check-node j select one group of uniform quantization parameter indicated with more bits;Again based on the uniform quantization parameter, successively gradually reduce the quantized value number of three kinds of variables, until obtaining the quantized value of number needed for meeting three kinds of variables of decoding performance requirement, these quantized values form three kinds of different non-uniform quantizing parameter sets, are quantified respectively to three kinds of variate-values using three kinds of non-uniform quantizing parameters.Non-uniform quantizing method proposed by the invention can use 3 bit quantization bit wides, reach the decoding performance under 6 bit quantization bit wide of uniform quantization method, effectively reduces resource needed for decoder hardware is realized, has preferable application value.
Description
Technical field
The invention belongs to communication channel decoding technique fields, in particular to based on low density parity check code and product decoding fortune
The non-uniform quantizing method for the value that counts.
Background technique
With the fast development of China's high-resolution earth observation systems, satellite earth observation precision is increasingly promoted, and is needed
The data volume of transmission is skyrocketed through, star data transmission rate demand promoted by hundreds of megabits per second to multiple gigabit per second.
And be limited by the power of radio-frequency devices on star, star number pass the transmission rate of link and promoted by hundreds of megabits per second to thousands of million
Bits per second is faced with huge challenge, needs to use the channel coding of high-gain to alleviate the office of junctor level surplus anxiety
Face.
In existing channel coding method, low density parity check code (Low-Density Parity-Check code,
Hereinafter referred to as LDPC code) there is coding gain to have error code flat low, simultaneously closest to the Shannon Coding Theory limit, simultaneously
Row decodes the advantages that being suitble to high-speed transfer, has strong application demand in satellite data transmission field.However, existing be based on FPGA core
The ldpc decoder that piece design is realized, rate is only hundreds of megabits per second, to realize number in the limited fpga chip of resource
The high-performance ldpc decoder of giga bits per second rate needs further to explore the side for reducing LDPC realization of decoding complexity
Method.
Efficient iterative numerical quantization is the key that realize the design of low complex degree ldpc decoder.LDPC code commonly decodes
Method is mainly based upon confidence spread and product decoding algorithm, and with product decoding algorithm in hardware realization, to every one-step decoding
Operation result will carry out numerical quantization processing.The process of numerical quantization be exactly high-precision numerical value set be mapped to one it is discrete
The Finite Number set of range value, with there is limit bit number to represent several quantized values, the maximum quantized value of numerical value is taken as maximum quantization
Value, positive maximum quantization value and negative minimum quantization value represent quantizing range, and the difference of two neighboring quantized value is quantized interval, quantization
Processing represents whole numerical value in an interval range with each quantized value, this section is taken as the corresponding quantization area of the quantized value
Between, the corresponding quantized interval of maximum quantization value is taken as maximum quantization section.The uniform quantization method amount of referring to generallyd use at present
Change and be spaced identical quantization method, hardware realization complexity is higher, and research high effective quantization method can be provided with lower hardware
Source cost reaches better performance, for solving message transmission rate and integrity problem in communication system, has extremely important
Meaning.
LDPC code is a kind of block code, and using supersparsity random matrix as check matrix, matrix structure can be with two-way
Tanner grid chart indicates.Fig. 1 show a two-way Tanner figure, and each black lattice represents check matrix H in figure
Corresponding LDPC code member composition one is equivalent to the pact of check code by the constraint relationship of middle a line, i.e., every all nonzero elements of row
Beam, this verification contextual definition are a check-node j.Each black small circle represents the column in check matrix H in figure
All nonzero elements of the constraint relationship, i.e. each column all correspond to the same symbol in LDPC code word, they are together with receiving symbol
The constraint for being equivalent to duplication code is constituted, this constraint relationship is defined as a bit node i.It is every in check matrix H
A nonzero element had not only participated in the constraint relationship of check-node but also had participated in the constraint relationship of bit node, was mapped as in Tanner figure
Connect " tie line " of two kinds of nodes.The output external information of check-node and bit node passes through the " connection between both nodes
Line " transmitting carries out mutual feedback iterative decoding.During decoding, the decoding of bit node is carried out first.Each bit node connects
The correspondence symbol Soft Inform ation for carrying out self-channel is received, that is, corresponds to symbol and takes the probability of " 1 " divided by the value for taking the probability of " 0 " to obtain
Natural logrithm is taken again, while each bit node also receives the check-node output external information being connected with the node, utilizes this
After a little information are decoded according to bit node the constraint relationship, bit node output external information is transmitted to connected school
Test node, each check-node is decoded using the external information that transmitting comes according to check-node the constraint relationship, then will decoding knot
Fruit exports to the bit node being connected.External information is transmitted between bit node and check-node back and forth along " tie line ",
Complete the overall process of decoding.
LDPC code and product decoding algorithm implementation process based on uniform quantization method are as follows:
1) to decoding initialization:
Receiving end receives the sequence of real numbers R=[R for carrying out self-channel0,R1,…,RN-1], wherein N is LDPC code code word size,
LDPC code each code element is adjusted to the form of log-likelihood ratio, the input Soft Inform ation as the LDPC code symbol:
LLR(RiThe σ of)=2/2Ri,1≤i≤N
Wherein σ2For interchannel noise standard variance.(Q, Q are carried out to calculated each code element Soft Inform ationf) at uniform quantization
Reason:
Wherein Q takes quantizing bit number, QfTake Soft Inform ation fractional part quantizing bit number, x code fetch member Soft Inform ation numerical value.
2) bit node decoding is carried out:
Each bit node i is using come each LDPC code symbol Soft Inform ation of self-channel and the verification section being attached thereto
Point output external information is decoded, and the relationship with output for "and" is inputted:
Wherein Col [i] indicates the set of nonzero element position in the i-th column in check matrix H, ri (n)(j) indicate that n-th changes
In generation, is output to the external information of bit node i from check-node j.If first time iteration, the number of iterations n=1, under primary condition
Check-node output external information is set as zero, i.e. ri (1)(j)=0.Calculated each bit node output external information is carried out equal
Even quantification treatment, processing method is identical as step 1), and wherein Q takes quantizing bit number, QfBit node is taken to export external information decimal
Part quantizing bit number, x take bit node to export external information numerical value.
3) check-node decoding is carried out:
Each check-node j is decoded using connected bit node output external information, is inputted and is exported
For the relationship of " product ":
Wherein Row [j] indicates that the set of check matrix H jth row nonzero element position, function phi (x) are defined as
Uniform quantization processing, processing method and step 1) phase are carried out to calculated each check-node output external information
Together, wherein Q takes quantizing bit number, QfCheck-node is taken to export external information fractional part quantizing bit number, x takes check-node to export
External information numerical value.
4) decoding result hard decision is obtained into hard decision result:
It is in each bit node that all input external informations and corresponding LDPC code symbol are soft after the completion of nth iteration
Information sums to obtain decoding result:
Hard decision is carried out to the decoding result and obtains hard decision result:
5) if hard decision result meets verification relationshipIt is then legal-code, is made
Finally to decode result output, step 2) is otherwise jumped to, iteration is increased as n=n+1, progress next round iteration;It is set if reaching
Fixed maximum number of iterations, then stop iteration, and last time iteration hard decision result is exported.
Above-mentioned LDPC code and product decoding algorithm receive sequence Soft Inform ation LLR (R for receiving end when realizing on hardwarei)、
Bit node i is output to the external information of check-node j in each iterative processCheck-node j is output to bit node i
External information ri (n)(j), it requires to do its numerical value quantification treatment and carries out finite-precision arithmetic.The numerical quantization generallyd use
Method is the uniform quantization method of equal quantized intervals, uniform quantization parameter (Q, Qf) indicate, Q indicates uniform quantization digit, QfTable
Show quantization decimal digits, if numerical value to be quantified exceeds quantizing rangeThen distinguished according to real number symbol to be quantified
Take quantized valueIf quantized value takes without departing from quantizing rangeWherein x indicate to
Quantized value, [] indicate the maximum integer for being no more than its value.Uniform quantization method is to reach and performance similar in non-quantization operations
Need larger quantization range and lesser quantized interval, it is desirable that all quantized values are indicated with more bit number, hard
Part needs to occupy more resources on realizing.
During decoding, Soft Inform ation, check-node output external information and the bit node output that receiving end receives are outer
Its numerical values recited of information is in different sections, and traditional (Q, Qf) uniform quantization, to the numerical value of Soft Inform ation and two kinds of external informations
Quantification treatment uses same quantizing range and quantized interval, i.e., uses unified uniform quantization to Soft Inform ation and two kinds of external informations
Parameter.Since bit node output external information is with the increase of the number of iterations, rapidly converge near maximum quantization value, check-node
External information is exported with the increase of the number of iterations, is slowly converged near maximum quantization value, sequence Soft Inform ation numerical value is received and is decoding
It remains unchanged in the process, so in given the number of iterations, each quantization of Soft Inform ation and two kinds of external informations for uniform quantization
Section susceptibility is different, and Soft Inform ation is high to specific quantized interval susceptibility, bit node export external information to it is small Liang Hua area
Between susceptibility it is low, and check-node output external information is low to big other quantized interval susceptibility.To guarantee decoding performance, generally require
More quantized value is selected, indicates that the quantizing bit number of these quantized values is more, wastes hardware resource, and quantized value is excessively few then tight
Ghost image rings decoding performance, therefore is difficult to find compromise between quantizing bit number and decoding performance.
Summary of the invention
It is an object of the invention to overcome the uniform quantization side of existing low density parity check code and product decoding operation numerical value
The deficiency of method proposes the non-uniform quantizing method of a kind of new low density parity check code (LDPC) and product decoding operation numerical value,
The performance with higher bit uniform quantization parameter similar can be reached with less quantizing bit number, substantially reduce hardware resource, had
There is stronger engineering practical value.
The non-uniform quantizing method of low density parity check code proposed by the present invention and product decoding operation numerical value, feature exist
In being output to the external information of bit node to symbol Soft Inform ation, from check-node, be output to the outer of check-node from bit node
Different non-uniform quantizing parameters is respectively adopted in these three variables of information;First to the Soft Inform ation of i-th of symbol of LDPC code, comparison
The external information that special node i is output to the external information of check-node j, is output to bit node i to check-node j selected compares more one group
The uniform quantization parameter that spy indicates;Again based on the uniform quantization parameter, the quantization of three kinds of variables is successively gradually reduced
It is worth number, finally obtains the quantized value of number needed for meeting three kinds of variables of decoding performance requirement, these quantized values forms three kinds
The set of different non-uniform quantizing parameters, using the different non-uniform quantizing parameter respectively to three kinds of variate-value amounts of progress
Change
This method specifically includes the following steps:
1) the lowest evenness quantizing bit number for meeting decoding performance requirement is determined by emulating, obtain the even amount of Soft Inform ation
Change parameter (QL,QfL), bit node exports the uniform quantization parameter (Q of external informationq,Qfq), check-node exports the equal of external information
Even quantization parameter (Qr,Qfr), wherein QL,Qq,QrRespectively represent the quantizing bit number of three kinds of information, QfL,Qfq,QfrRespectively represent three
The quantizing bit number of the fractional part of kind information;
2) the uniform quantization parameter (Q obtained based on step 1)L,QfL),(Qq,Qfq) and (Qr,Qfr), successively gradually subtract respectively
Few LLR (Ri)、And ri (n)(j) quantized value number is to arrivingIt is a, use QnonIndicate non-uniform quantizing value bit number,
QnonValue is no more than QL,Qq,QrThe positive integer of minimum value;Elder generation is since maximum quantization value and passage capacity emulation successively determines
The of each variable..., 1 non-uniform quantizing value, specific implementation step is as follows:
2.1) determination receives each code element Soft Inform ation LLR (Ri)A non-uniform quantizing value;The soft letter of symbol
Cease uniform quantization parameter (QL,QfL) in willThe positive real number of a forward quantizations value is divided intoA forward direction quantized interval,The negative realpartition of a negative quantified value atA negative sense quantized interval, positive smallest quantization interval and negative maximum quantization
The quantized value in region is zero between section;Maximum quantization section is merged with time big quantized interval for the first time, by combined quantization area
Between middle smallest quantization interval median asA non-uniform quantizing value;With the non-uniform quantizing value, and do not merge
The former uniform quantization value and variable of quantized intervalAnd ri (n)(j) (Qq,Qfq) and (Qr,Qfr) uniform quantization parameter,
Performance simulation is carried out, if the bit error rate is in same quantity when ternary is all made of uniform quantization parameter in decoding performance and step 1)
Grade is then cancelled current quantized interval and is merged;Secondly the non-uniform quantizing value maximum three quantized intervals merged, again into
Row performance simulation, if the bit error rate cancels this in the same order of magnitude when decoding performance and ternary are all made of uniform quantization parameter
Secondary quantized interval merges, and so on, the number for merging quantized interval is gradually increased, until the N-1 times, maximum N number of quantization area
Between merge after, the bit error rate increases an order of magnitude or more when simulation performance is all made of uniform quantization parameter than ternary, then cancels
Current quantized interval merges, using the median of smallest quantization interval when the N-2 time merging as the of Soft Inform ationA amount
Change value, N are positive integer;
2.2) in Soft Inform ation LLR (Ri) using the determining non-uniform quantizing value of step 2.1), external information ri (n)(j) it uses
(Qr,Qfr) on uniform quantization parameter basis, external information is determined using the same procedure with step 2.1)?A non-uniform quantizing value and does not merge quantized interval original uniform quantization value, step 2.1) with the non-uniform quantizing value
Determining LLR (Ri) non-uniform quantizing value and variable ri (n)(j) (Qr,Qfr) uniform quantization parameter, performance simulation is carried out, with
This analogizes, and carries out performance simulation to the non-uniform quantizing parameter obtained after each merge, obtains the median of smallest quantization interval
As?A quantized value;
2.3) external information r is determined using the same procedure with step 2.1)i (n)(j)A quantized value, with this
Non-uniform quantizing value, and does not merge quantized interval original uniform quantization value and LLR (Ri) using the non-homogeneous of step 2.1) determination
Quantized value,The non-uniform quantizing value determined using step 2.2) carries out performance simulation, minimum quantization area when obtaining merging
Between median as ri (n)(j)A quantized value;
2.4) to fixed three variables L LR (Ri)、ri (n)(j)A non-uniform quantizing value is big
It is small to carry out 1/Q respectivelyfL、1/QfqAnd 1/QfrThe more of the non-uniform quantizing value of three variables are combined into after the increasing or decreasing of multiple
A combination carries out performance simulation to each combination, finally, all simulation performances are compared, selects three when optimal performance
A variable non-uniform quantizing value is as newly determining LLR (Ri)、ri (n)(j)A non-uniform quantizing value;
2.5) step 2.1) -2.4 is repeated), determine LLR (Ri)、And ri (n)(j)..., 1 non-uniform quantizing value;
3) by the LLR (R after determinationi)、ri (n)(j)A non-uniform quantizing value obtain three kinds it is non-
Even quantized value set
The non-uniform quantizing value gathered three kinds emulates, as final non-homogeneous amount if decoding performance meets the requirements
Change value, the return step 2 if decoding performance is undesirable) redefine the non-uniform quantizing value of ternary.
The features of the present invention and effect:
Method of the invention is based on during low density parity check code and product decoding operation, and Soft Inform ation and bit node are defeated
The characteristics of external information and check-node output external information are in different numberical ranges out, uses the uniform quantization indicated with more bits
Based on parameter, these three variables are respectively adopted not in the method for successively gradually reducing Soft Inform ation and external information quantized value number
Same quantized value and quantized interval carries out quantification treatment, finally obtains the non-uniform quantizing parameter of lowest bit number expression.It is this
The non-uniform quantizing method spy different with quantized interval locating for two kinds of external informations during decoding for LDPC code symbol Soft Inform ation
The low uniform quantization section of three kinds of variable susceptibilitys is merged or is deleted respectively, can reduce quantized value to a greater extent by point
Number can reach close using each quantization bit to greatest extent to reduce the quantizing bit number for being used to indicate quantized value
The decoding performance for the uniform quantization parameter that former more bits indicate.On the uniform quantization parameter basis that initial more bits indicate, often
One wheel search is all successively by Soft Inform ation, bit node output external information, the maximum several uniform quantizations of check-node output external information
Section merges, and removal influences the corresponding quantized value of insensitive quantized interval to performance, until obtaining QnonBit indicates non-equal
Even quantization parameter.
The present invention can reach the even amount indicated better than more bits with the non-uniform quantizing parameter that less bit indicates
The decoding performance for changing parameter, reaches compromise on decoding performance and complexity, hence it is evident that reduces hardware resource, has stronger application
Prospect.
Detailed description of the invention
Fig. 1 is the two-way Tanner expression figure of LDPC code.
Fig. 2 is the flow chart for realizing non-uniform quantizing method of the invention.
Specific embodiment
Non-uniform quantizing method proposed by the present invention based on low density parity check code and product decoding operation numerical value combines
Detailed description are as follows for embodiment:
Non-uniform quantizing method such as Fig. 2 proposed by the present invention based on low density parity check code and product decoding operation numerical value
Shown, this method specifically includes and is output to the external information of bit node, defeated from bit node to symbol Soft Inform ation, from check-node
The external information of check-node these three variables are arrived out, and different non-uniform quantizing parameters is respectively adopted;The choosing of non-uniform quantizing parameter
It selects based on the uniform quantization parameter that more bits indicate, successively gradually reduces the quantized value number of three kinds of variables, finally
The quantized value of number needed for obtaining three kinds of variables, the collection of these quantized values are combined into non-uniform quantizing parameter, using it is different it is non-
Even quantization parameter respectively quantifies three kinds of variate-values.
This method specifically includes the following steps:
1) the uniform quantization parameter of selection LDPC and product decoding operation numerical value:
The lowest evenness quantizing bit number for meeting decoding performance requirement is determined by emulating, and obtains the uniform quantization of Soft Inform ation
Parameter (QL,QfL), bit node exports the uniform quantization parameter (Q of external informationq,Qfq), check-node exports the uniform of external information
Quantization parameter (Qr,Qfr), wherein QL,Qq,QrRespectively represent the quantizing bit number of three kinds of information, QfL,Qfq,QfrRespectively represent three kinds
The quantizing bit number of the fractional part of information;Specific implementation are as follows:
Sequence of real numbers R=[the R that receiving end receives0,R1,R2,…,RN-1], i-th, i=0,1 ... i ..., N-1 code
First Soft Inform ation LLR (Ri) quantized value are as follows:
The external information of check-node j is output to during decoding to nth iteration bit node iQuantized value are as follows:
The external information r of bit node i is output to during decoding to nth iteration check-node ji (n)(j) quantized value
Are as follows:
Wherein [] indicates the maximum integer for being no more than its value;
2) the uniform quantization parameter (Q obtained based on step 1)L,QfL),(Qq,Qfq) and (Qr,Qfr), successively gradually subtract respectively
Few LLR (Ri)、And ri (n)(j) quantized value number is extremelyIt is a, use QnonIndicate non-uniform quantizing value bit number,
QnonValue is no more than QL,Qq,QrThe positive integer of minimum value;(due to positive and negative sense quantized value be it is full symmetric, theA quantized value is zero, thus the non-uniform quantizing parameter for each variable it needs to be determined thatA quantized value), pass through
Performance simulation successively determines the of each variable since maximum quantization value..., 1 non-homogeneous amount
Change value, the specific steps are as follows:
2.1) determination receives each code element Soft Inform ation LLR (Ri)A non-uniform quantizing value;The soft letter of symbol
Cease uniform quantization parameter (QL,QfL) in willThe positive real number of a forward quantizations value is divided intoA forward direction quantized interval,The negative realpartition of a negative quantified value atA negative sense quantized interval, positive smallest quantization interval and negative maximum quantization area
Between between the quantized value in region be zero;Maximum quantization section is merged with time big quantized interval for the first time, by combined quantized interval
The median of middle smallest quantization interval is asA non-uniform quantizing value;With the non-uniform quantizing value, with the amount of merging
Change the former uniform quantization value and variable in sectionAnd ri (n)(j) (Qq,Qfq) and (Qr,Qfr) uniform quantization parameter, into
Row performance simulation, if the bit error rate is in the same order of magnitude when ternary is all made of uniform quantization parameter in decoding performance and step 1),
Then cancel current quantized interval to merge;Secondly the non-uniform quantizing value maximum three quantized intervals merged, re-starts
Performance simulation is cancelled current quantized interval and is merged if decoding performance and ternary are all made of close when uniform quantization parameter, with
This analogizes, and gradually increases the number for merging quantized interval, until the N-1 times, after maximum N number of quantized interval merges, simulation performance
The bit error rate increases an order of magnitude or more when being all made of uniform quantization parameter than ternary, then cancels current quantized interval and merge,
When the N-2 times is merged the median of smallest quantization interval as Soft Inform ation theA quantized value, N are positive integer;
2.2) in Soft Inform ation LLR (Ri) using the determining non-uniform quantizing value of step 2.1), external information ri (n)(j) it uses
(Qr,Qfr) on uniform quantization parameter basis, external information is determined using the same procedure with step 2.1)?A quantized value, specifically: maximum quantization section is merged with time big quantized interval for the first time, by combined quantized interval
The median of middle smallest quantization interval is asA non-uniform quantizing value, with the non-uniform quantizing value, with the amount of merging
Change section original uniform quantization value, the LLR (R that step 2.1) determinesi) non-uniform quantizing value and variable ri (n)(j) (Qr,Qfr)
Uniform quantization parameter carries out performance simulation, if ternary is all made of uniform quantization parameter in same decoding performance and step 1)
The bit error rate is then cancelled current quantized interval and is merged in the same order of magnitude;Secondly maximum three quantized intervals are merged, is re-started
Performance simulation, if the bit error rate is cancelled current in the same order of magnitude when decoding performance and ternary are all made of uniform quantization parameter
Quantized interval merges, and so on, until the N-1 times, after maximum N number of quantized interval merges, simulation performance is adopted than ternary
Increase an order of magnitude or more with bit error rate when uniform quantization parameter, then cancels current quantized interval and merge, the N-2 times is merged
When smallest quantization interval median conduct?A quantized value;
2.3) external information r is determined using the same procedure with step 2.1)i (n)(j)A quantized value, specifically
Are as follows: maximum quantization section is merged with time big quantized interval for the first time, by the centre of smallest quantization interval in combined quantized interval
Value is as theA non-uniform quantizing value and does not merge quantized interval original uniform quantization value with the non-uniform quantizing value,
And LLR (Ri) using step 2.1) determine non-uniform quantizing value,The non-uniform quantizing determined using step 2.2)
Value carries out performance simulation, if the bit error rate is in same number when ternary is all made of uniform quantization parameter in decoding performance and step 1)
Magnitude is then cancelled current quantized interval and is merged, and secondly merges maximum three quantized intervals, performance simulation is re-started, if translating
The bit error rate is then cancelled current quantized interval and is merged in the same order of magnitude when code performance and ternary are all made of uniform quantization parameter,
And so on, until the N-1 times, after maximum N number of quantized interval merges, simulation performance is all made of uniform quantization parameter than ternary
When the bit error rate increase an order of magnitude or more, then cancel current quantized interval and merge, by smallest quantization interval when the N-2 times merging
Median as ri (n)(j)A quantized value;
24) three variables L LR (R are being determinedi)、ri (n)(j)After a non-uniform quantizing value, to this
The of three variablesThe size of a non-uniform quantizing value carries out 1/Q respectivelyfL、1/QfqAnd 1/QfrThe increase of multiple or
It is combined into multiple combinations of the non-uniform quantizing value of three variables after reduction, performance simulation is carried out to each combination, finally, by institute
There is simulation performance to compare, selects three variable non-uniform quantizing values when optimal performance as finally determining LLR (Ri)、ri (n)(j)A non-uniform quantizing value;
Specific embodiment can are as follows: the first two variable (for example, LLR (R firsti)、) theIt is a non-homogeneous
Quantized value remains unchanged, third variable (for example, variable ri (n)(j))A non-uniform quantizing value increases 1/Qfr's
1-4 times of size carries out performance simulation;Secondly the first two variable theA non-uniform quantizing value remains unchanged, by third
VariableA non-uniform quantizing value reduces 1/Qfr1-4 times of size, then carry out performance simulation;Again, first change
Flow controlA non-uniform quantizing value remains unchanged, by second variableA non-uniform quantizing value increases by 1/
Qfq1-4 times of size, third variableA non-uniform quantizing value increases or decreases 1/Qfr1-4 times of size, into
Row performance simulation;Again, first variableA non-uniform quantizing value remains unchanged, by second variableA non-uniform quantizing value reduces 1/Qfq1-4 times of size, third variableA non-uniform quantizing value increases
Add deduct few 1/Qfr1-4 times of size, carry out performance simulation;Again, first variableA non-uniform quantizing value
Increase 1/QfL1-4 times of size, by second variableA non-uniform quantizing value increases or decreases 1/Qfq1-4 times
Size, third variableA non-uniform quantizing value increases or decreases 1/Qfr1-4 times of size, it is imitative to carry out performance
Very;Again, first variableA non-uniform quantizing value reduces 1/QfL1-4 times of size, by second variableA non-uniform quantizing value increases or decreases 1/Qfq1-4 times of size, third variableIt is a non-homogeneous
Quantized value increases or decreases 1/Qfr1-4 times of size, carry out performance simulation;Finally, all simulation performances are compared, select
Ternary non-uniform quantizing value when optimal performance is as updated LLR (Ri)、ri (n)(j)It is a non-
Uniform quantization value.Put again or contracting uniform quantization parameter under the 1-4 times of size and combining of quantized interval be determined, select performance most
Excellent parameter combination is as finally determining LLR (Ri)、ri (n)(j)A non-uniform quantizing value.
2.5) step 2.1) -2.4 is repeated), determine LLR (Ri)、And ri (n)(j)..., 1 non-uniform quantizing value;(in each non-uniform quantizing value for determining ternary, it will
Performance simulation is carried out on previous step fixed non-uniform quantizing value basis, when proceeding to determining Soft Inform ation LLR (Ri), it is outer
InformationAnd external information ri (n)(j) when the 1st quantized value, determine that the 1st quantization is joined merging maximum several quantized intervals
Number when, the quantized value for not merging quantized interval takes 0, using the median of the smallest quantization interval merged in best merging method as
1st non-uniform quantizing value of ternary.)
3) in LLR (Ri)、ri (n)(j)A non-uniform quantizing value has just obtained non-equal after all determining
Even quantization parameter (Qnonun-L,Qnonun-q,Qnonun-r), wherein
Represent LLR (Ri) non-uniform quantizing
Value set,
It representsNon-uniform quantizing
Value set,
Represent ri (n)(j) non-uniform quantizing
Value set,
LLR(Ri) non-uniform quantizing value are as follows:
Non-uniform quantizing value are as follows:
ri (n)(j) non-uniform quantizing value are as follows:
It is emulated for this non-uniform quantizing value, final non-uniform quantizing value is used as if decoding performance meets the requirements,
The return step 2 if decoding performance is undesirable) redefine the non-uniform quantizing value of ternary.
The present invention proposes the implementation of a kind of non-uniform quantizing method of low density parity check code and long-pending decoding operation numerical value
Example is described as follows:
The LDPC code word parameter of the present embodiment are as follows: code length 15330, information bit length 12775, code rate 5/6, the present embodiment
Select the non-uniform quantizing parameter indicated for three bits of above-mentioned code word and product decoding operation numerical value, specific implementation step
It is as follows:
1) the uniform quantization parameter of selection LDPC and product decoding operation numerical value:
The lowest evenness quantizing bit number for meeting decoding performance requirement is determined by emulating, and obtains the uniform quantization of Soft Inform ation
Parameter (QL,QfL) it is (6,1), bit node exports the uniform quantization parameter (Q of external informationq,Qfq) it is (6,2), check-node is defeated
Uniform quantization parameter (the Q of external information outr,Qfr) it is (6,5), wherein QL=6, Qq=6, Qr=6 respectively represent the amount of three kinds of information
Change bit number, QfL=1, Qfq=2, Qfr=5 respectively represent the quantizing bit number of the fractional part of three kinds of information.It is implemented as
Sequence of real numbers R=[the R received for receiving end0,R1,R2,…,R15329], i-th, i=0,1 ..., i ..., 15329 codes
Soft Inform ation LLR (the R of memberi), the external information of check-node j is output to bit node i during decodingVerification is saved
Point j is output to the external information r of bit node ii (n)(j) quantify are as follows:
Wherein [] indicates the maximum integer for being no more than its value, for the quantization x=LLR (R of Soft Inform ationi), Qf=1, for
The quantization x=q of variable node output external informationj(i), Qf=2, for the quantization x=r of check-node output external informationi(j), Qf
=5, n represent the number of iterations.
2) the uniform quantization parameter (6,1), (6,2) and (6,5) obtained based on step 1), successively gradually reduces LLR respectively
(Ri)、 And ri (n)(j) quantized value number uses Q to 7non=3 indicate non-uniform quantizing value bit number;First from most
Big quantized value starts and passage capacity emulates the 3rd, 2,1 non-uniform quantizing value for successively determining each variable;Specific implementation step
It is rapid as follows:
2.1) determination receives each code element Soft Inform ation LLR (Ri) the 3rd non-uniform quantizing value;Symbol Soft Inform ation is uniform
The positive real number of 31 forward quantizations values is divided into 32 positive quantized intervals in quantization parameter (6,1), 31 negative quantified values it is negative
Realpartition is at 32 negative sense quantized intervals, and the quantized value in region is between positive smallest quantization interval and negative maximum quantization section
Zero;The the 32nd, 31 quantized interval is merged for the first time, regard the median 15 of the 31st quantized interval as the 3rd non-uniform quantizing
Value;With the non-uniform quantizing value 15, and the former uniform quantization value [0,0.5,1.0,1.5 ..., 14.5] for not merging quantized interval,
And variableAnd ri (n)(j) (6,2) and (6,5) uniform quantization parameter carries out performance simulation, decoding performance and step
1) bit error rate is cancelled current quantized interval and is merged in the same order of magnitude when ternary is all made of uniform quantization parameter in;Secondly will
32nd, 31,30 quantized interval is merged, regard the median 14.5 of the 30th quantized interval as the 3rd non-uniform quantizing value;
With the non-uniform quantizing value 14.5, and the former uniform quantization value [0,0.5,1.0,1.5 ..., 14] for not merging quantized interval, with
And variableAnd ri (n)(j) (6,2) and (6,5) uniform quantization parameter, re-starts performance simulation, decoding performance and three
The bit error rate is cancelled current quantized interval and is merged in the same order of magnitude when variable is all made of uniform quantization parameter, and so on, gradually
Increase the number for merging quantized interval, until the 21st time, after maximum 22 quantized intervals merge, simulation performance is adopted than ternary
Increase an order of magnitude or more with bit error rate when uniform quantization parameter, then cancels current quantized interval and merge, the 20th time is merged
When the 12nd quantized interval 3rd quantized value of the median 5.5 as Soft Inform ation;
2.2) in Soft Inform ation LLR (Ri) using the determining non-uniform quantizing value of step 2.1), external information ri (n)(j) use (6,
5) on uniform quantization parameter basis, external information is determined using the same procedure with step 2.1)The 3rd quantized value, tool
Body are as follows: the 32nd, 31 quantized interval is merged for the first time, regard the median 7.5 of the 31st quantized interval as the 3rd non-homogeneous amount
Change value and does not merge quantized interval original uniform quantization value [0,0.25,0.5,0.75 ..., 7.25] with the non-uniform quantizing value,
LLR (the R that step 2.1) determinesi) non-uniform quantizing value and variable ri (n)(j) (6,5) uniform quantization parameter carries out performance
The bit error rate is cancelled current in the same order of magnitude when ternary is all made of uniform quantization parameter in emulation, decoding performance and step 1)
Quantized interval merges;Secondly the 32nd, 31,30 quantized interval is merged, by 7.25 conduct of median of the 30th quantized interval
3rd non-uniform quantizing value, with the non-uniform quantizing value, with do not merge quantized interval original uniform quantization value [0,0.25,0.5,
0.75 ..., 7], the LLR (R that step 2.1) determinesi) non-uniform quantizing value and variable ri (n)(j) (6,5) uniform quantization ginseng
Number re-starts performance simulation, and the bit error rate takes in the same order of magnitude when decoding performance and ternary are all made of uniform quantization parameter
The current quantized interval that disappears merges, and so on, until the 12nd time, after maximum 13 quantized intervals merge, simulation performance becomes than three
The bit error rate increases an order of magnitude or more when amount is all made of uniform quantization parameter, then cancels current quantized interval and merge, by the 11st
5.0 conduct of median of 21st quantized interval when secondary mergingThe 3rd quantized value;
2.3) external information r is determined using the same procedure with step 2.1)i (n)(j) the 3rd quantized value, specifically: it is first
It is secondary to merge the 32nd, 31 quantized interval, it regard the median 0.9375 of the 31st quantized interval as the 3rd non-uniform quantizing
Value, with the non-uniform quantizing value, with do not merge quantized interval original uniform quantization value [0,0.03125,0.0625,0.09375 ...,
0.90625] and LLR (Ri) using step 2.1) determine non-uniform quantizing value,It is determined using step 2.2) non-
Uniform quantization value carries out performance simulation, and the bit error rate exists when ternary is all made of uniform quantization parameter in decoding performance and step 1)
The same order of magnitude is cancelled current quantized interval and is merged, secondly merges the 32nd, 31,30 quantized interval, by the 30th quantization area
Between median 0.90625 be used as the 3rd non-uniform quantizing value, it is former with quantized interval is not merged with the non-uniform quantizing value
Even quantized value [0,0.03125,0.0625,0.09375 ..., 0.875] and LLR (Ri) using the non-of step 2.1) determination
Uniform quantization value,The non-uniform quantizing value determined using step 2.2), re-starts performance simulation, decoding performance and three
The bit error rate is cancelled current quantized interval and is merged in the same order of magnitude when variable is all made of uniform quantization parameter, and so on, until
3rd time, after maximum 4 quantized intervals merge, the bit error rate increases one when simulation performance is all made of uniform quantization parameter than ternary
More than a order of magnitude, then cancels current quantized interval and merge, the median of the 30th quantized interval when the 2nd time is merged
0.90625 is used as ri (n)(j) the 3rd quantized value;
2.4) three variables L LR (R are being determinedi)、ri (n)(j) after the 3rd non-uniform quantizing value, this three are become
After the size of 3rd non-uniform quantizing value of amount carries out 0.5,0.25 and 0.03125 the increasing or decreasing of 1-4 times of quantity respectively
Multiple combinations of the non-uniform quantizing value of three variables are combined into, performance simulation are carried out to each combination, finally, by all emulation
Performance compares, and selects three variable non-uniform quantizing values when optimal performance as finally determining LLR (Ri)、ri (n)(j) the 3rd non-uniform quantizing value;
Specific embodiment can are as follows: the first two variables L LR (R firsti) and3rd non-uniform quantizing value is kept not
Become, third variable ri (n)(j) the 3rd non-uniform quantizing value increases by 0.03125 1-4 times of size, carries out performance simulation;Secondly
The 3rd non-uniform quantizing value of the first two variable remains unchanged, and the 3rd non-uniform quantizing value of third variable is reduced 0.03125
1-4 times of size, then carry out performance simulation;Again, the 3rd non-uniform quantizing value of first variable remains unchanged, by second
The 3rd non-uniform quantizing value of variable increases by 0.25 1-4 times of size, and the 3rd non-uniform quantizing value increasing of third variable adds deduct
Few 0.03125 1-4 times of size carries out performance simulation;Again, the 3rd non-uniform quantizing value of first variable remains unchanged, will
The 3rd non-uniform quantizing value of second variable reduces by 0.25 1-4 times of size, and the 3rd non-uniform quantizing value of third variable increases
Add deduct few 0.03125 1-4 times of size, carry out performance simulation;Again, the 3rd non-uniform quantizing value of first variable increases
The 3rd non-uniform quantizing value of second variable is increased or decreased 0.25 1-4 times of size, third by 0.5 1-4 times of size
The 3rd non-uniform quantizing value of variable increases or decreases 0.03125 1-4 times of size, carries out performance simulation;Again, first change
3 non-uniform quantizing values of flow control reduce by 0.5 1-4 times of size, and the 3rd non-uniform quantizing value of second variable is increased or decreased
0.25 1-4 times of size, the 3rd non-uniform quantizing value of third variable increase or decrease 0.03125 1-4 times of size, carry out
Performance simulation;Finally, all simulation performances are compared, select ternary non-uniform quantizing value when optimal performance as more
LLR (R after newi)、ri (n)(j) the 3rd non-uniform quantizing value, result 7.0,6.0 and 0.96875.
2.5) step 2.1) -2.4 is repeated), determine LLR (Ri)、And ri (n)(j) the 2nd non-uniform quantizing value be
4.0,3.25 and 0.25;Repeat step 2.1) -2.4), determine LLR (Ri)、And ri (n)(j) the 1st non-uniform quantizing
Value, when the maximum several quantized intervals of merging determine the 1st quantization parameter, the quantized value for not merging quantized interval takes 0, determines
The 1st non-uniform quantizing value of ternary is 2.5,1.75 and 0.03125.
3) in LLR (Ri)、ri (n)(j) 3 non-uniform quantizing values have just obtained non-uniform quantizing after all determining
Parameter (Qnonun-L,Qnonun-q,Qnonun-r), wherein Qnonun-L=[- 7.0, -4.0, -2.5,0,2.5,4.0,7.0] represent LLR
(Ri) non-uniform quantizing value set, Qnonun-q=[- 6.0, -3.25, -1.75,0,1.75,3.25,6.0] represent's
Non-uniform quantizing value set, Qnonun-r=[- 0.96875, -0.25, -0.03125,0,0.03125,0.25,0.96875] represents
ri (n)(j) non-uniform quantizing value set, LLR (Ri) non-uniform quantizing value are as follows:
Non-uniform quantizing value are as follows:
ri (n)(j) non-uniform quantizing value are as follows:
It is emulated for this non-uniform quantizing value, in Eb/N0The bit error rate reaches 9.860 × 10 when for 3.7dB-8, accidentally
Code rate reaches 1 × 10-7When do not occur that error code is flat, and decoding performance meets the requirements, as final 3 bit non-uniform quantizing parameter.
Performance of the different quantization schemes of table 1LDPC code under awgn channel
Table 2LDPC code difference quantization scheme FPGA realizes resource comparison (FPGA model: XC7VX485T)
Claims (2)
1. a kind of non-uniform quantizing method of low density parity check code and product decoding operation numerical value, which is characterized in that symbol
Soft Inform ation, the external information that bit node is output to from check-node, be output to from bit node check-node external information this three
Different non-uniform quantizing parameters is respectively adopted in kind variable;First to the Soft Inform ation of i-th of symbol of LDPC code, to bit node i
The selected expression of bit more than one group of external information for being output to the external information of check-node j, being output to bit node i to check-node j
Uniform quantization parameter;Again based on the uniform quantization parameter, the quantized value number of three kinds of variables is successively gradually reduced,
Finally obtain the quantized value of number needed for meeting three kinds of variables of decoding performance requirement, these quantized values form three kinds it is different non-
The set of uniform quantization parameter respectively quantifies three kinds of variate-values using the different non-uniform quantizing parameter;
This method specifically includes the following steps:
1) the lowest evenness quantizing bit number for meeting decoding performance requirement is determined by emulating, obtain the uniform quantization ginseng of Soft Inform ation
Number (QL,QfL), bit node exports the uniform quantization parameter (Q of external informationq,Qfq), check-node exports the even amount of external information
Change parameter (Qr,Qfr), wherein QL,Qq,QrRespectively represent the quantizing bit number of three kinds of information, QfL,Qfq,QfrRespectively represent three kinds of letters
The quantizing bit number of the fractional part of breath;
2) the uniform quantization parameter (Q obtained based on step 1)L,QfL),(Qq,Qfq) and (Qr,Qfr), successively gradually reduce respectively
LLR(Ri)、And ri (n)(j) quantized value number is to arrivingIt is a, wherein LLR (Ri) indicate i-th of symbol of LDPC code
Soft Inform ation,Bit node i is output to the external information of check-node j, r during nth iterationi (n)(j) n-th is indicated
Check-node j is output to the external information of bit node i in secondary iterative process, uses QnonIndicate non-uniform quantizing value bit number, Qnon
Value is no more than QL,Qq,QrThe positive integer of minimum value;First simultaneously passage capacity emulation successively determines often since maximum quantization value
The of a variable..., 1 non-uniform quantizing value;
3) by the LLR (R after determinationi)、ri (n)(j)A non-uniform quantizing value obtains three kinds of non-homogeneous amounts
Change value set, the non-uniform quantizing value gathered three kinds emulates, as final non-homogeneous if decoding performance meets the requirements
Quantized value, the return step 2 if decoding performance is undesirable) redefine the non-uniform quantizing value of ternary.
2. the non-uniform quantizing method of low density parity check code as described in claim 1 and product decoding operation numerical value, feature
It is, the step 2) specific implementation step is as follows:
2.1) determination receives each code element Soft Inform ation LLR (Ri)A non-uniform quantizing value;Symbol Soft Inform ation is equal
Even quantization parameter (QL,QfL) in willThe positive real number of a forward quantizations value is divided intoA forward direction quantized interval,
The negative realpartition of a negative quantified value atA negative sense quantized interval, between positive smallest quantization interval and negative maximum quantization section
The quantized value in region is zero;Maximum quantization section is merged with time big quantized interval for the first time, it will be minimum in combined quantized interval
The median of quantized interval is asA non-uniform quantizing value;With the non-uniform quantizing value, quantify area with not merging
Between former uniform quantization value and variableAnd ri (n)(j) (Qq,Qfq) and (Qr,Qfr) uniform quantization parameter progress performance
Emulation, if the bit error rate is cancelled in the same order of magnitude when ternary is all made of uniform quantization parameter in decoding performance and step 1)
Current quantized interval merges;Secondly the non-uniform quantizing value maximum three quantized intervals merged, it is imitative to re-start performance
Very, if the bit error rate cancels current quantization area in the same order of magnitude when decoding performance and ternary are all made of uniform quantization parameter
Between merge, and so on, gradually increase the number for merging quantized interval, until the N-1 time, after maximum N number of quantized interval merging,
The bit error rate increases an order of magnitude or more when simulation performance is all made of uniform quantization parameter than ternary, then cancels current quantization area
Between merge, when the N-2 times is merged the median of smallest quantization interval as Soft Inform ation theA quantized value, N are positive
Integer;
2.2) in Soft Inform ation LLR (Ri) using the determining non-uniform quantizing value of step 2.1), external information ri (n)(j) (Q is usedr,Qfr)
On uniform quantization parameter basis, external information is determined using the same procedure with step 2.1)?It is a non-equal
Even quantized value and does not merge quantized interval original uniform quantization value, the LLR (R that step 2.1) determines with the non-uniform quantizing valuei)
Non-uniform quantizing value and variable ri (n)(j) (Qr,Qfr) uniform quantization parameter, performance simulation is carried out, and so on, to every
The non-uniform quantizing parameter obtained after secondary merging carries out performance simulation, obtains the median conduct of smallest quantization interval's
TheA quantized value;
2.3) external information r is determined using the same procedure with step 2.1)i (n)(j)A quantized value, it is non-with this
Even quantized value, and does not merge quantized interval original uniform quantization value and LLR (Ri) using the determining non-uniform quantizing of step 2.1)
Value,Performance simulation is carried out using the non-uniform quantizing value that step 2.2) determines, smallest quantization interval when obtaining merging
Median is as ri (n)(j)A quantized value;
2.4) to fixed three variables L LR (Ri)、ri (n)(j)A non-uniform quantizing value size point
It carry out not 1/QfL、1/QfqAnd 1/QfrMultiple groups of the non-uniform quantizing value of three variables are combined into after the increasing or decreasing of multiple
It closes, performance simulation is carried out to each combination, finally, all simulation performances are compared, select three changes when optimal performance
Non-uniform quantizing value is measured as newly determining LLR (Ri)、ri (n)(j)A non-uniform quantizing value;
2.5) step 2.1) -2.4 is repeated), determine LLR (Ri)、And ri (n)(j)
..., 1 non-uniform quantizing value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610282279.6A CN105978578B (en) | 2016-04-29 | 2016-04-29 | The non-uniform quantizing method of low density parity check code and product decoding operation numerical value |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610282279.6A CN105978578B (en) | 2016-04-29 | 2016-04-29 | The non-uniform quantizing method of low density parity check code and product decoding operation numerical value |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105978578A CN105978578A (en) | 2016-09-28 |
CN105978578B true CN105978578B (en) | 2019-03-29 |
Family
ID=56993529
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610282279.6A Active CN105978578B (en) | 2016-04-29 | 2016-04-29 | The non-uniform quantizing method of low density parity check code and product decoding operation numerical value |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105978578B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110024297B (en) * | 2016-10-14 | 2023-07-18 | 马维尔亚洲私人有限公司 | System and method for log likelihood ratio based dynamic preprocessing selection scheme in low density parity check decoder |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101120530A (en) * | 2004-12-22 | 2008-02-06 | 高通股份有限公司 | Systems and methods that utilize a capacity-based signal-to-noise ratio to predict and improve mobile communication |
CN101854179A (en) * | 2010-05-26 | 2010-10-06 | 厦门大学 | 5bit quantization method applied to LDPC decoding |
WO2013117076A1 (en) * | 2012-02-07 | 2013-08-15 | 中兴通讯股份有限公司 | Method and system for iterative decoding |
CN103905064A (en) * | 2014-04-23 | 2014-07-02 | 湖南国科微电子有限公司 | Method for quantization of LDPC encoder |
CN103916134A (en) * | 2014-03-24 | 2014-07-09 | 清华大学 | Low-density parity check code aliasing and decoding method and multi-core collaborative aliasing decoder |
CN103957015A (en) * | 2014-05-12 | 2014-07-30 | 福州大学 | Nonuniform quantizing coding method used for decoding LDPC code and application of method in decoder |
-
2016
- 2016-04-29 CN CN201610282279.6A patent/CN105978578B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101120530A (en) * | 2004-12-22 | 2008-02-06 | 高通股份有限公司 | Systems and methods that utilize a capacity-based signal-to-noise ratio to predict and improve mobile communication |
CN101854179A (en) * | 2010-05-26 | 2010-10-06 | 厦门大学 | 5bit quantization method applied to LDPC decoding |
WO2013117076A1 (en) * | 2012-02-07 | 2013-08-15 | 中兴通讯股份有限公司 | Method and system for iterative decoding |
CN103916134A (en) * | 2014-03-24 | 2014-07-09 | 清华大学 | Low-density parity check code aliasing and decoding method and multi-core collaborative aliasing decoder |
CN103905064A (en) * | 2014-04-23 | 2014-07-02 | 湖南国科微电子有限公司 | Method for quantization of LDPC encoder |
CN103957015A (en) * | 2014-05-12 | 2014-07-30 | 福州大学 | Nonuniform quantizing coding method used for decoding LDPC code and application of method in decoder |
Non-Patent Citations (2)
Title |
---|
LDPC码改进的量化自适应偏移最小和算法;董同昕等;《计算机工程与应用》;20141231;全文 |
Performance Optimization of Generalized Low-Density Parity-Check Codes with EXIT Chart Method for Data Transmission over Partial-Band Jamming Channels;Qi Li,et al.;《2014 International Conference on Information and Communication Technology Convergence (ICTC)》;20141231;全文 |
Also Published As
Publication number | Publication date |
---|---|
CN105978578A (en) | 2016-09-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yuan et al. | Low-latency successive-cancellation list decoders for polar codes with multibit decision | |
Zhang et al. | Shuffled belief propagation decoding | |
CN106998208B (en) | Code word construction method of variable length Polar code | |
Lacruz et al. | One minimum only trellis decoder for non-binary low-density parity-check codes | |
CN103929210B (en) | Hard decision decoding method based on genetic algorithm and neural network | |
Buchberger et al. | Pruning neural belief propagation decoders | |
CN105763203A (en) | Multi-element LDPC code decoding method based on hard reliability information | |
CN104393877B (en) | Irregular LDPC codes linear programming interpretation method based on weighting | |
CN110233628B (en) | Self-adaptive belief propagation list decoding method for polarization code | |
CN109067408A (en) | A kind of design method of protograph LDPC code | |
CN107196737B (en) | SCMA decoding method based on message passing algorithm | |
CN109586733A (en) | A kind of LDPC-BCH interpretation method based on graphics processor | |
CN110166171A (en) | Multielement LDPC code compensates high-performance decoding scheme based on the segmented of EMS | |
Yang et al. | Nonlinear programming approaches to decoding low-density parity-check codes | |
CN106209305A (en) | A kind of fountain codes interpretation method under access channel | |
CN105978578B (en) | The non-uniform quantizing method of low density parity check code and product decoding operation numerical value | |
CN101854179B (en) | 5bit quantization method applied to LDPC decoding | |
CN101895375B (en) | Decoding system of low-density parity-check code | |
CN110958025B (en) | Short frame length coding and decoding method based on superposition | |
Han et al. | Simplified multi-bit SC list decoding for polar codes | |
CN117914446B (en) | Decoding method and system for algebraic code | |
CN108173550A (en) | A kind of general simplified method suitable for nearly all NB-LDPC interpretation methods | |
CN107124186B (en) | Two-stage decoding method of LDPC code based on grid complexity | |
CN109756234A (en) | Long code constructing method in a kind of LDPC of the high code rate based on matroid theory | |
Huang et al. | A Markov chain model for edge memories in stochastic decoding of LDPC 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |