A multi-standard efficient column-layered LDPC decoder for Software Defined Radio on GPUs
Rongchun Li, Jie Zhou, Yong Dou, Song Guo, Dan Zou, Shi Wang · 2013
In this paper, we propose a multi-standard high-throughput column-layered (CL) low-density parity-check (LDPC) decoder for Software-Defined Radio (SDR) on a Graphics Processing Unit (GPU) platform. Multiple columns in the sub-matrix of quasi-cyclic LDPC (QC-LDPC) code are parallel performed inside a block, while multiple codewords are simultaneously decoded among many blocks on the GPU. Several optimization methods are employed to enhance the throughput, such as the compressed matrix structure, memory optimization, codeword packing scheme, two-dimension thread configuration and asynchronous data transfer. The experiment shows that our decoder has low bit error ratio and the peak throughput is 712Mbps, which is about two orders of magnitude faster than that of CPU implementation and comparable to the dedicated hardware solutions. Compared to the existing fastest GPU-based implementation, the presented decoder can achieve a performance improvement of 3.0x times.