DocumentCode :
3079185
Title :
A low power VLSI design paradigm for iterative decoders
Author :
Elassal, Mahmoud ; Baker, Abu ; Bayoumi, Magdy
Author_Institution :
Center for Adv. Comput. Studies, Louisiana Univ., Lafayette, LA, USA
fYear :
2005
fDate :
2-4 Nov. 2005
Firstpage :
272
Lastpage :
277
Abstract :
In this paper we present low power maximum a posteriori (MAP) decoder architectures using dual supply voltages. The architecture leverages an application specific integrated circuits (ASIC) structure, where the architecture components that require a higher performance are powered from a high supply voltages VddH, and the less demanding components are powered from a low supply voltage VddL. Salient features of this architecture include: (a) high level of parallelism, (b) reduced power consumption without affecting the architecture performance, and (c) a tradeoff between the decoding time delay and the number of state metric banks, branch metric banks, and state metric update kernels respectively. The power consumption reduction of the dual-supply voltage over the single-supply voltage has been estimated and the memory access frequencies as well. The proposed architecture achieves approximate 35-40% power reduction from the single-supply architecture.
Keywords :
VLSI; application specific integrated circuits; graph theory; iterative decoding; maximum likelihood decoding; power consumption; trellis codes; ASIC structure; application specific integrated circuits; architecture performance; branch metric banks; dual supply voltages; iterative decoders; low power VLSI design paradigm; low power maximum a posteriori decoder; memory access frequency; power consumption reduction; state metric banks; state metric update kernels; time delay decoding; Application specific integrated circuits; Computer architecture; Concatenated codes; Energy consumption; Frequency estimation; Hardware; Iterative decoding; Parallel processing; Very large scale integration; Voltage;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Signal Processing Systems Design and Implementation, 2005. IEEE Workshop on
ISSN :
1520-6130
Print_ISBN :
0-7803-9333-3
Type :
conf
DOI :
10.1109/SIPS.2005.1579878
Filename :
1579878
Link To Document :
بازگشت