Research On Decoding Performance Of LDPC Codes With Low Error Floor

Low-density parity-check(LDPC) codes has come to be one of the best codingtechnologies because of their low-complicity iterative decoding algorithm and nearShannon limit performance. Due to the advantages of LDPC codes, their applications inreliable communications have received great interests in academia and IT field. LDPCcodes have been used in optical network and digital storage.This paper mainly focus on error floor of LDPC codes and trapping sets eliminatingalgorithm. Based on the application of importance sample(IS) method to identifytrapping sets of LDPC codes, a new method to eliminate tapping sets is proposed. Someresults are obtained and summarized as follow:1. Theory of LDPC codes and its Tanner graphs are briefly illuminated, the beliefpropagation decoding algorithm is analyzed in detail. Error floor is studied based onsome properties of LDPC codes, such as cycles and connectivity, and the harmfulnessof trapping sets is explained.2. Principle of IS method and the feasibility of its applicaition on searchingtapping sets are thoroughly analyzed, the algorithms of IS method adapt to identifytrapping sets in regular and irregular LDPC codes are given separately, the veracity andefficiency of this method are proved by emulation results.3. A novel method to eliminate small trapping sets in irregular LDPC codes isproposed in this paper. By adding several new check nodes and having their edgesconnected to small trapping sets for eliminating, this algorithm can eliminate smalltrapping sets in the original code, and reduce the error floors significantly. Emulationresults reveal that this method is efficient and convenient to implement.