CN109831216A - Polarization code SBP decoder based on G-Matrix verification - Google Patents
Polarization code SBP decoder based on G-Matrix verification Download PDFInfo
- Publication number
- CN109831216A CN109831216A CN201910051819.3A CN201910051819A CN109831216A CN 109831216 A CN109831216 A CN 109831216A CN 201910051819 A CN201910051819 A CN 201910051819A CN 109831216 A CN109831216 A CN 109831216A
- Authority
- CN
- China
- Prior art keywords
- information bit
- information
- decoding
- vector
- segmentation
- 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.)
- Pending
Links
Landscapes
- Error Detection And Correction (AREA)
Abstract
The present invention provides the efficient SBP decoders that a kind of segmentation increases information bit R information initial value, are made of information bit reliability sorting module, information bit segment processing module, G-Matrix correction verification module etc..The method that it chooses information bit based on Gaussian approximation, sorts from large to small processing for information bit reliability and obtains index vectorThen to vectorIt is divided into n subvector according to segmentation radix μIt is sequentially increased vectorThe initial value of interior information bit R information finally decodes again and seeks decoding optimization using G-Matrix verification.The present invention is considered as information bit and is converted into freeze position to a certain extent, i.e., effective information is constant can but achieve the effect that and decoding performance is promoted when code rate reduces.The present invention is simple and effective, only scans for, changes and verifies when decoding error, substantially increases error correcting capability, and the effect of optimization of BP decoding performance is obvious, and as n=16, bit error rate performance promotes 0.25dB.
Description
Technical field
The present invention relates to technical field of electronic communication more particularly to polarization code channel decoding technical field, in particular to divide
The initial value of Duan Zeng great information bit R information verifies optimization BP decoding by G-Matrix to design segmentation SBP (Segment
Belief Propagation) decoder.
Background technique
Arikan proposes channel-polarization concept within 2009, by being combined to same channel, then to synthesis after
Channel is split, and the channel capacity a part formed after splitting levels off to 0, and another part levels off to 1, forms the two poles of the earth point
The phenomenon that change.Polarization code is a kind of novel channel coding proposed based on channel-polarization, uniquely can achieve Shannon so far
The channel coding method of the limit has milestone formula meaning in coding theory.Polarization code coding method, which is broadly divided into, continuously to be deleted
Except (Successive Cancellation, SC) decoding and two kinds of belief propagation (Belief Propagation, BP) decoding,
SC decoding deletes the related information of its redundancy after being estimated by turn source signal, and brings into estimated value as prior information
Decoding operation later.However for time-limited polarization code, subchannel is not perfact polarization, causes SC decoding algorithm in
The error-correcting performance effect in short code field is undesirable, therefore gradually elimination list (Successive is introduced on the basis of SC
Cancellation List, SCL) decoding algorithm further improves the performance of SC algorithm, but due to increasing mulitpath
L, brought computation complexity and decoding latency are consequently increased.For this purpose, scholars have further delivered some improved SCL
Version such as simplifies gradually to cancel list (SSCL) and quickly simplify and gradually offsets (FSSCL), and having a definite purpose is to improve handling capacity
With reduction decoding latency.Although SC decoder passes through a series of improvement, for decoder handling capacity and decoding delay, with BP
Decoding algorithm is compared to still unsatisfactory.BP decoder is the structure of a complete parallel, possess higher decoding throughput and
Shorter decoding latency has wider application in real life.However, the error-correcting performance of BP decoding is less desirable,
It is far from reaching theoretical upper limit value.
Therefore, scholar proposes that subchannel capacities are low, error-prone information bit and LDPC code cascade, in conjunction with LDPC
The advantage of code and Polar code improves decoding performance；Some scholars use the SCAN of similar SC algorithm structure to scan decoding algorithm, add
Fast decoding convergence rate；There are also scholars to propose the BPL decoding algorithm based on displacement BP factor graph structure.The present invention is based on BPL to translate
The thought of code algorithm, a kind of method for proposing initial value for selectively increasing information bit R information, then utilizes the school G-Matrix
It tests and seeks to decode optimal solution.It being experimentally confirmed, this method makes BP decoding error correcting capability reach a higher level, with
SCL decoding algorithm is more competitive.
Summary of the invention
The present invention is in order to solve the problems, such as that BP decoder performance is insufficient, at the beginning of proposing a kind of segmentation increase information bit R information
The high-performance SBP decoder of initial value.Coding/decoding system is communicated as shown in Figure 1, wherein decoder is by information bit reliability sequence mould
The composition such as block, information bit segment processing module, G-Matrix correction verification module.
In BP decoding, all bit informations are all constantly iteratively to transmit soft-decision LLR by left and right on factor graph
Information is freezed information and is updated from left to right, and channel information updates from right to left.The information for freezing position carrying is referred to as R information,
Do not carry effective information due to freezing position, freeze position R information initial value setting be positive it is infinite, at the beginning of the R information of information bit
Initial value is set as 0.Fig. 2 is the BP factor graph of code length N=8, in the left end of BP factor graph, R1,iRepresent the initial value of R information；?
The right end L of factor graphM+1,i(M=log2N, N are code length, and K is information bit length, i ∈ { 1,2 ..., N }) represent channel input terminal
The initial value of log-likelihood ratio information is found according to experiment, when correctly increasing information bit R information initial value, decoding performance meeting
It gets a promotion.Information bit reliability sorting module mainly sorts from large to small the reliability of N bit informations, first with height
The method that this approximation chooses information bit calculates the seat N channel capacity C=[C1C2C3,...,CN], by C according to descending from big to small
Processing can obtain vector C '=[C '1C′2C′3,...,C′N], while introducing a new vectorSo that C '
=C (λ (i)), the more forward subchannel capacities that represent of sorting are bigger, and channel is more stable, while when meaning to transmit information more
Accurately and reliably.Enter information bit segment processing module after sequence is completed, in vectorIn take before K be used as information bit vectorBy vectorIt is divided into n subvector according to segmentation radix μWherein μ≤2M-4, n=1,2,
3..., vectorLength x with segmentation radix μ increase and reduce, effective range x is as shown in table 1, x be not less than 24, j=
1,2,3,...,M-4.It is sequentially increased vectorThe initial value of interior information bit R information, finally decoding enters G-Matrix again
Correction verification module meets the i.e. output decoding of verification, still verifies when n reaches maximum value and unsuccessfully exports original BP decoding result.Fig. 3
Be code length be 1024, code rate be 0.5 under different signal-to-noise ratio frame error rate and the bit error rate curve, as n=16, BP decoding performance
0.25dB is promoted, experiments have shown that this method has feasibility.
The present invention is segmented the R information for changing information bit according to chopping rule, seeks the excellent of decoding using G-Matrix verification
Change.The rule of segmentation defers to segmentation radix from small to large, and the sequence of reliability from high to low successively carries out, and segmentation radix is smaller,
Effective range x is bigger, and the range of change is wider, and the performance of promotion is more obvious, but is located at the wrong ratio that range decoding occurs
Special number is consequently increased, it is therefore desirable to be increased segmentation radix, be reduced search range, reduce number of error bits.
There are certain bit error rates during actual decoding, but when there is decoding error, in general
The code word of error only accounts for whole fraction, so when you increase information bit R information initial value and are precisely correctly decoding, this
When these information bits be considered as to freeze position, i.e., effective information is constant can but reach and decoding performance is promoted when code rate reduces
Effect.In addition, although the initial value for changing information bit R information is to vectorIt is segmented, but due to polarization code factor graph
The relationship of structure, the information bit of change are randomly dispersed in each node, and being closely connected between code bit and position of polarizing, and are had
Certain probability can correct the information bit of original mistake, therefore decoding meets G-Matrix verification, obtains correctly decoding result.
Detailed description of the invention
Fig. 1 is communication coding/decoding system structural schematic diagram；
Fig. 2 is BP factor structure figure (N=8) schematic diagram；
Fig. 3 is code length 1024, frame error rate FER and error rate BER curve graph (n_max of the code rate 0.5 under different signal-to-noise ratio
For segmentation sum)；
Fig. 4 is decoding flow chart (t is the number of iterations).
Specific embodiment
Present invention is further described in detail with reference to the accompanying drawing, and following embodiment facilitates the understanding of the present invention,
It is relatively good application example, but is not construed as limiting thereof.
Step 1: the seat N channel capacity C=[C is calculated using the method that Gaussian approximation chooses information bit1C2C3,...,
CN], C is handled according to descending from big to small can obtain vector C '=[C '1C′2C′3,...,C′N], while introducing a new vectorSo that C '=C (λ (i)).
Step 2: BP iterative decoding terminates calculation using formula (1) G-Matrix that Bo Yuan in existing literature is mentioned in advance
Method, wherein G is generator matrix,WithBP factor graph left and right ends Hard decision decoding is respectively represented, can be obtained according to formula (2)
It arrives, M=log2N, i ∈ [1, N].
Step 3: when being unsatisfactory for G-Matrix verification, illustrate that this time mistake occurs in decoding, in vectorIn take before K make
For information bit vectorBy vectorIt is divided into n subvector according to segmentation radix μWherein μ
≤ K, n=1,2,3..., effective range x is as shown in table 1, then changes vector according to formula (3)Interior information bit R information
Initial value
Step 4: the initial value of information bit R informationAfter update, the nodal information in factor graph is initialized, is retained
Original freezes a R information initial value and channel output L information, guarantees to update independently every time in this way, then decode again, Zhi Daoman
Sufficient G-Matrix verification.
Step 5: original BP decoding is unsuccessfully exported as a result, guaranteeing performance not if still verified when n reaches maximum value
It suffers a loss.
Under the different segmentation radix μ of table 1Effective range x
The above embodiments do not limit the invention in any form, all using similar structure of the invention, method and its similar
Variation pattern technical solution obtained, in protection scope of the present invention.
Claims (4)
1. it is initial to propose a kind of segmentation increase information bit R information in order to solve the problems, such as that BP decoder performance is insufficient by the present invention
The efficient SBP decoder of value, it is characterised in that an information bit reliability sorting module and an information bit segment processing module.
2. information bit reliability sorting module according to claim 1, it is characterised in that Gaussian approximation chooses information bit
Method calculates the seat N channel capacity C=[C1C2C3,...,CN], C is handled according to descending from big to small can obtain vector C '=[C1′
C2′C3′,...,C′N], while introducing a new vectorSo that C '=C (λ (i)).
3. information bit segment processing module according to claim 1, it is characterised in that by vectorAccording to radix μ points of segmentation
At n subvectorWherein μ≤2M-4, n=1,2,3..., vectorEffective range x it is as shown in table 1, x be not less than 24,
J=1,2,3 ..., M-4.Chopping rule defers to segmentation radix from small to large, and the sequence of reliability from high to low successively carries out, point
Section radix is smaller, and effective range x is bigger, and the range of change is wider, and the performance of promotion is more obvious, but is located at the range and decodes out
Existing number of error bits is consequently increased, it is therefore desirable to be increased segmentation radix, be reduced search range, reduce number of error bits.
4. information bit segment processing module according to claim 1, it is characterised in that change vector according to following equation
The initial value of interior information bit R informationIfInterior information bit is precisely correctly decoding, is considered as information
Position, which is converted into, freezes position, i.e., effective information is constant can but achieve the effect that be promoted with decoding performance when code rate reduction.
Under the different segmentation radix μ of table 1Effective range x
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910051819.3A CN109831216A (en) | 2019-01-21 | 2019-01-21 | Polarization code SBP decoder based on G-Matrix verification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910051819.3A CN109831216A (en) | 2019-01-21 | 2019-01-21 | Polarization code SBP decoder based on G-Matrix verification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109831216A true CN109831216A (en) | 2019-05-31 |
Family
ID=66861036
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910051819.3A Pending CN109831216A (en) | 2019-01-21 | 2019-01-21 | Polarization code SBP decoder based on G-Matrix verification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109831216A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110278002A (en) * | 2019-06-19 | 2019-09-24 | 东南大学 | Polarization code belief propagation list decoding method based on bit reversal |
| CN112152752A (en) * | 2019-06-27 | 2020-12-29 | 中兴通讯股份有限公司 | Decoding processing method and device |
| WO2021007751A1 (en) * | 2019-07-15 | 2021-01-21 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for improved belief propagation based decoding |
-
2019
- 2019-01-21 CN CN201910051819.3A patent/CN109831216A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110278002A (en) * | 2019-06-19 | 2019-09-24 | 东南大学 | Polarization code belief propagation list decoding method based on bit reversal |
| CN110278002B (en) * | 2019-06-19 | 2023-01-17 | 东南大学 | Bit-flipping-based polarization code belief propagation list decoding method |
| CN112152752A (en) * | 2019-06-27 | 2020-12-29 | 中兴通讯股份有限公司 | Decoding processing method and device |
| CN112152752B (en) * | 2019-06-27 | 2023-09-29 | 中兴通讯股份有限公司 | Decoding processing method and device |
| WO2021007751A1 (en) * | 2019-07-15 | 2021-01-21 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for improved belief propagation based decoding |
| US11886418B2 (en) | 2019-07-15 | 2024-01-30 | Telefonaktiebolaget Lm Ericsson (Publ) | Method and apparatus for improved belief propagation based decoding |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108566213B (en) | Serial offset list bit flipping decoding method of polarization code | |
| RU2595542C2 (en) | Device and method for transmitting and receiving data in communication/broadcasting system | |
| CN109257148B (en) | Polarization code BP decoding method based on Gaussian approximate threshold judgment | |
| CN101656541B (en) | Coding method and device of RS codes | |
| CN107919874B (en) | Syndrome computation basic check node processing unit, method and computer program | |
| CN109831216A (en) | Polarization code SBP decoder based on G-Matrix verification | |
| WO2005096509A1 (en) | Multi-threshold message passing decoding of low-density parity check codes | |
| CN110278002A (en) | Polarization code belief propagation list decoding method based on bit reversal | |
| US10637510B2 (en) | Methods and devices for error correcting codes decoding | |
| CN110999093B (en) | Hybrid architecture for check node processing for extended minimum sum (EMS) decoding of non-binary LDPC codes | |
| CN104158550A (en) | A method for constructing a code-rate-compatible protograph LDPC code based on deep space communication environment | |
| US11133827B2 (en) | Simplified, presorted, syndrome-based, extended min-sum (EMS) decoding of non-binary LDPC codes | |
| Feng et al. | A novel high-rate polar-staircase coding scheme | |
| Gamage et al. | Low latency decoder for short blocklength polar codes | |
| CN111034055A (en) | Simplified check node processing in non-binary LDPC decoders | |
| WO2019075988A1 (en) | Decoding method and device | |
| CN116614142A (en) | Combined decoding method based on BPL decoding and OSD decoding | |
| KR20090012189A (en) | Apparatus and method for decoding using performance enhancement algorithm for ldpc codes with scaling based min-sum iterative decoding | |
| Zhang et al. | BER evaluation based SCFlip algorithm for polar codes decoding | |
| Cai et al. | An improved simplified soft cancellation decoding algorithm for polar codes based on frozen bit check | |
| US20220337269A1 (en) | Block code encoding and decoding methods, and apparatus therefor | |
| CN114221664A (en) | Low-complexity polar code simplified soft elimination list decoder and decoding method | |
| Zhilin et al. | High-rate codes for high-reliability data transmission | |
| CN113556135A (en) | Polarization code belief propagation bit flipping decoding method based on frozen flipping list | |
| US10382066B2 (en) | Method and apparatus for decoding three-dimensional turbo product code based on crossing layers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190531 |