Title :
Low-power dual quantization-domain decoding for LDPC codes
Author :
Abu-Surra, Shadi ; Pisek, Eran ; Henige, Thomas ; Rajagopal, Sridhar
Author_Institution :
Samsung Res. America-Dallas, Samsung Electron., Richardson, TX, USA
Abstract :
In this paper we propose a new dual quantization-domain LDPC decoder, which requires only 3-bit messages between the check nodes and the variable nodes. To reduce complexity and save power, check nodes processing is entirely in the 3-bit domain. However, to avoid loss in performance, the variable nodes processing is in a higher bit precision domain. A non-linear mapping is used to map the messages from one quantization domain to the other. Simulations and hardware evaluation of the proposed decoder showed greater than 50% reduction in power consumption with only 0.1 dB (0.2 dB) loss in bit-error-performance when compared to a reference 5-bits scaled Min-Sum decoder over the AWGN (fading) channel.
Keywords :
computational complexity; electronic messaging; error statistics; parity check codes; power consumption; quantisation (signal); telecommunication power management; 3-bit domain; AWGN; LDPC codes; bit-error-performance; low-power dual quantization-domain decoding; min-sum decoder; nonlinear mapping; power consumption reduction; AWGN channels; Bit error rate; Decoding; Hardware; Iterative decoding; Quantization (signal); LDPC; decoder architecture; iterative decoder; layered scheduling; low-power; non-linear mapping; quantization;
Conference_Titel :
Global Communications Conference (GLOBECOM), 2014 IEEE
Conference_Location :
Austin, TX
DOI :
10.1109/GLOCOM.2014.7037290