CN105306073A

CN105306073A - Method for improving performance of LDPC (Low Density Parity Check) decoder by utilizing given information

Info

Publication number: CN105306073A
Application number: CN201510740472.5A
Authority: CN
Inventors: 邢新景; 黄立忠; 于攀; 陈烈
Original assignee: Hangzhou National Chip Science & Technology Co Ltd
Current assignee: Hangzhou National Chip Science & Technology Co Ltd
Priority date: 2015-11-04
Filing date: 2015-11-04
Publication date: 2016-02-03

Abstract

The invention relates to a method for improving performance of an LDPC (Low Density Parity Check) decoder by utilizing given information, which is suitable for the field of digital television reception technologies. The conventional decoder design method equally treats an information bit and a check bit which are included in an LDPC code word, and does not utilize the structural feature of the LDPC code word and the structural feature of the LDPC code. According to the method, information carried by a signaling domain of the LDPC code word is taken as the given information, and the highest degree of confidence is endowed in each iteration process, and the improved degree of confidence information is continuously transmitted to other associated information bits and check bits in the iteration process, so that the speed for iterative convergence is accelerated; and furthermore, the transmission process for the degree of confidence information of the given information to intervene the degree of confidence continuously is adjusted in the iterative process under the condition that the signal quality is critical, the accelerating convergence of the degree of confidence information is guided, the error correcting capability of the decoder is improved, and some code words which cannot be corrected by the existing decoding method are corrected.

Description

A kind of method utilizing Given information to improve ldpc decoder performance

Technical field

The invention belongs to Digital TV Receiving technical field, relate to a kind of method utilizing Given information to improve ldpc decoder performance.

Background technology

Because LDPC (low density parity check code) has good error-correcting performance, be widely used in field of digital television transmission in recent years gradually as channel error correction encoding.The error-correcting performance that LDPC is good comes from iterative decoding, the information bit of LDPC code word in iterative decoding process and the confidence information of check digit constantly feed back transmission, through continuous iteration, confidence information restrains gradually, along with the convergence decode results of confidence information also restrains gradually, and when restraining, ldpc decoder can obtain correct decode results.The convergence rate of decode results determines the iterations needed for decode procedure.When the signal quality that receiver receives from channel is good time, through minority several times iterative decoding result can restrain, but when the signal quality that receiver receives from channel is not good time, the situation that no matter still can not restrain through how many times iteration may be there is, just there will be decoding failure in this case.Actual ldpc decoder circuit is owing to being subject to the restriction of the factor such as circuit scale, circuit power consumption, always wish in design process can reduce the iterations reached needed for correct decoding as far as possible, and better restrained result when equal iterations.

Existing ldpc decoder workflow as shown in Figure 1.When iteration is initial according to the confidence information Vi (i=1 of code word computing information position and check digit before the decoding received, 2, N. wherein N is the length of LDPC code word), then often carry out the computing of a LDPC iterative decoding just whether to restrain decode results and once judge, if decode results restrains, terminate decoding Output rusults, otherwise carry out the LDPC iterative decoding computing of next round.

The information bit comprised in LDPC code word, check digit are not added putting on an equal footing of differentiation by existing decoder design method, do not utilize the architectural feature of LDPC code word in digital television transmission system and the structural feature of LDPC code itself.The structure of LDPC code word as shown in Figure 2, be made up of LDPC information bit and LDPC check digit, wherein LDPC information bit is made up of signaling territory, payload, and signaling territory is made up of command transmitting and CRC, and whether the CRC in signaling territory is used for correct by cyclic redundancy calculating inspection command transmitting.The structural feature of described LDPC code word is that the signaling domain information protection of LDPC code to code word start-up portion is stronger, obtains correct result at first in an iterative process, if correct through CRC check confirmation, can be used in follow-up iterative process.

Summary of the invention

The object of the invention is for the deficiencies in the prior art, a kind of method utilizing Given information to improve ldpc decoder performance is provided.

The concrete steps of the inventive method are:

Step 1, according to the confidence information Vi of the information bit in code word compute codeword before the decoding that receives and check digit (i=1,2 ..., N.N is LDPC code word length); Computational methods adopt the computational methods in existing LDPC decoding;

Step 2, judge whether signaling domain information is Given information, if signaling domain information is known before decoding, signaling territory known markers S is put 1; If signaling domain information is unknown before decoding, known for signaling territory mark S is set to 0;

Step 3, carry out the computing of a LDPC iterative decoding; Computational methods adopt ripe prior art;

Step 4, judge whether decode results restrains: if decode results convergence, enter step 8; If decode results does not restrain, enter step 5; Described judges the determination methods of method for using in existing LDPC interpretation method whether decode results restrains;

Step 5, judge whether signaling territory known mark S is 1, if S=1, namely signaling territory is known, turns to step 6; If S ≠ 1, mark signaling territory is unknown, turns to step 7;

Step 6, by the confidence information Vi of corresponding for signaling territory position (i=1,2 ..., M.M is the length in signaling territory) be set to maximum and turn to step 3;

Step 7, do CRC check to corresponding position, signaling territory, and judge that whether check results is correct, if check results is correct, the acquisition that the information in signaling territory is correct is described, S is also put 1 by preservation signaling domain information, goes to step 6; If verify incorrect, signaling domain information is still unknown, turns to step 3 to carry out next round iteration; Described CRC check is cyclic redundancy check (CRC);

Step 8, end decoding Output rusults.

The inventive method propose on the basis of current ldpc decoder a kind of utilize Given information improve ldpc decoder performance method, the method using the information entrained by the signaling territory of LDPC code word as Given information, all the highest confidence level is given in each iterative process, these confidence information be enhanced constantly pass to other information bit be associated and check digit in an iterative process, thus accelerate the speed of iteration convergence; And when signal quality is critical, the transmittance process of confidence level is constantly intervened by the confidence information adjusting given data in an iterative process, guide the accelerating ated test of confidence information, improve the error correcting capability of decoder, can not the correct error correction of the code word of error correction by existing for part interpretation method.

Accompanying drawing explanation

Fig. 1 is existing LDPC interpretation method flow chart;

Fig. 2 is LDPC code word structural representation;

Fig. 3 be the present invention utilize Given information improve ldpc decoder performance method flow diagram.

Embodiment

Below in conjunction with accompanying drawing, with a specific embodiment, technical scheme of the present invention is described.Without loss of generality, the LDPC code adopting DVBS2 to use for convenience of description is in the following embodiments described, and corresponding parameter gets N=64800, M=80.

As shown in Figure 3, a kind of utilize Given information improve ldpc decoder performance method by the flow process in figure:

Step 1, according to the confidence information Vi of code word computing information position before the decoding that receives and check digit (i=1,2 ..., N.N=64800 is LDPC code word length); Computational methods adopt the log-likelihood calculations method in existing LDPC decoding;

Step 2, judge whether signaling domain information is Given information, if signaling domain information is known before decoding, signaling territory known markers S is put 1; If signaling domain information is unknown before decoding, known for signaling territory mark S is set to 0; Signaling domain information is unknown before decoding in the present embodiment, is set to 0 by known for signaling territory mark S;

Step 3, carry out the computing of a LDPC iterative decoding; Operation method adopts the conventional methods such as minimum and decoding;

Step 4, judge whether decode results restrains, if decode results convergence, enter step 8, if decode results does not restrain, enter step 5; Described judges the determination methods of method for using in existing LDPC interpretation method whether decode results restrains;

Step 6, by the confidence information Vi of corresponding for signaling territory position (i=1,2 ..., M.M=80 is the length in signaling territory) be set to maximum and turn to step 3;

Step 7, do CRC check to corresponding position, signaling territory, and judge that whether check results is correct, if check results is correct, the acquisition that the information in signaling territory is correct is described, S is also put 1 by preservation signaling domain information, goes to step 6; If verify incorrect, signaling domain information is still unknown, turns to step 3 to carry out next round iteration;

Step 8, end decoding Output rusults.

It should be understood that above-mentioned example is just to explanation of the present invention, instead of limitation of the present invention, any innovation and creation do not exceeded in spirit of the present invention, all fall within protection scope of the present invention.

Claims

1. utilize Given information to improve a method for ldpc decoder performance, it is characterized in that the concrete steps of the method are:

Information bit before the decoding that step 1, basis receive in code word compute codeword and the confidence information Vi of check digit, i=1,2 ..., N, N are LDPC code word length;

Step 3, carry out the computing of a LDPC iterative decoding;

Step 4, judge whether decode results restrains: if decode results convergence, enter step 8; If decode results does not restrain, enter step 5;

Step 6, the confidence information Vi of corresponding for signaling territory position is set to maximum and turns to step 3, i=1,2 ..., M, M are the length in signaling territory;

Step 7, do CRC check to corresponding position, signaling territory, and judge that whether check results is correct, if check results is correct, the acquisition that the information in signaling territory is correct is described, S is also put 1 by preservation signaling domain information, goes to step 6; If verify incorrect, signaling domain information is still unknown, goes to step 3 and carries out next round iteration;

Step 8, end decoding Output rusults.