Efficient GPU-based implementation for decoding non-binary LDPC codes with layered and flooding schedules

摘要

Nonbinary low-density parity-check (NB-LDPC) codes are excellent error correcting codes andrnoutperform their binary counterparts under the same code length. NB-LDPC decoders arernbased on Belief Propagation Algorithm, which demands intensive message-passing computation.rnRecently, to achieve both flexibility and good throughput performance, NB-LDPC decoders havernbeen ported from dedicated hardware solutions to multi/many-core systems. In this paper, wernpropose an FFT-based q-ary Sum-Product Algorithm (QSPA) decoding architecture forNB-LDPCrncodes with layered and flooding schedules on a graphics processing unit (GPU). To improve thernthroughput performance of the proposed decoder, four optimization methods are presented tornnot only accelerate the decoding kernel execution but also improve the data transfer efficiency.rnThe experiments are mainly accomplished on NVIDIA GTX580 and GTX Titan X. Throughputsrnup to 63 Mbps over GF(16) and 7.65 Mbps over GF(256) are achieved on GTX580 when executingrn5 layered decoding iterations. Throughputs can reach up to 139 Mbps over GF(16) andrn17 Mbps over GF(256) on GTX Titan X. Experimental results show that the speedups of therndecoding throughputs range from×1.7 to×16.8 by comparison with the existing FFT-basedQSPArndecoders on GPU.