Title :
An 847–955 Mb/s 342–397 mW Dual-Path Fully-Overlapped QC-LDPC Decoder for WiMAX System in 0.13
m CMOS
Author :
Xiang, Bo ; Bao, Dan ; Huang, Shuangqu ; Zeng, Xiaoyang
Author_Institution :
State Key Lab. of ASIC & Syst., Fudan Univ., Shanghai, China
fDate :
6/1/2011 12:00:00 AM
Abstract :
This paper presents a partially-parallel dual-path fully-overlapped QC-LDPC decoder for the WiMAX system. By adopting five techniques including symmetrical six-stage pipelining, block column and row interleaving, nonzero sub-matrix reordering, sum memory quad-partition and read-write bypass, the decoder continuously scans nonzero sub-matrices two by two in the block row-wise order without any memory access conflict. Two phases are fully overlapped with each other, and the check node updating phase always takes the latest sums from the previous variable node updating phase. The sum memory stores not only the posterior sums but also the prior messages, which saves 11,520 memory bits. It only takes 48-54 clock cycles for the decoder to finish one iteration. The read-write accesses to sum memories are reduced by 24.3%-48.8%. Fabricated in the SMIC 0.13 μ m CMOS process, the decoder occupies 4.84 mm 2 with core area of 3.03 mm2, attains 847-955 Mb/s at 214 MHz and 10 iterations, and consumes 342-397 mW at 1.2 V with power efficiency of 39-46 pJ per bit per iteration.
Keywords :
CMOS integrated circuits; WiMax; parity check codes; CMOS process; QC-LDPC decoder; WiMAX; bit rate 847 Mbit/s to 955 Mbit/s; power 342 mW to 397 mW; read write access; size 0.13 mum; sub-matrix reordering; voltage 1.2 V; Clocks; Decoding; Iterative decoding; Pipeline processing; Program processors; Registers; WiMAX; Block column and row interleaving; QC-LDPC codes; decoder architecture; dual-path; fully-overlapped; nonzero sub-matrix reordering; read-write bypass; symmetrical six-stage pipelining;
Journal_Title :
Solid-State Circuits, IEEE Journal of
DOI :
10.1109/JSSC.2011.2125030