Dong Zijian

Improved BP-OSD cascade decoding algorithm using limited likelihood information

In this work, we use restricted likelihood information of belief propagation (BP) algorithm to improve the accuracy of sorting in order statistic decoding (OSD) for low-density parity-check (LDPC) codes. The reliability information delivered from the last BP iteration to OSD algorithm is inaccurate because that the propagation information may be enlarged for the existence of girths, thence limitation on iterative information can been applied to improve the OSD decoding performance. Simulation results show that the proposed method is effective.

Selecting error patters based on symbol reliability for OSD algorithm

A new scheme of selecting error pattern is proposed. OSD algorithm processes certain MRIPs of a received sequence, performs bits flip among MRIPs, forms error patterns, and then generates candidate codeword. The bits flip is equal probability among MRIPs in OSD algorithms, which leads to too many candidates generated, and seriously affects the implement of algorithm. This paper proposes an improved algorithm which adopts unequal probability strategy utilizing the reliability of received symbols. The simulation results show that the proposed algorithm is effective. With little performance degradation, the computational complexity of algorithm can be reduced by 60% or more for one class of LDPC codes.

Reliability-based iterative decoding of LDPC codes using segmented accumulation of likelihood ratio

In this work, we present an improvement on the belief propagation (BP) algorithm concatenated serially with order statistic decoding (OSD) for low-density parity-check (LDPC) codes. We use the segmented accumulation of log-likelihood ratio (LLR) of variance nodes as the ordered information metric to reconstruct received sequence. This method can reduce the sensitivity of accumulated parameter selection to the fluctuation of performance, and further improve decoding performance.