Power reduction through adaptive data compression in Network-on-Chip architectures

Abstract

In this paper we present an adaptive compression technique that reduce the power consumption when compare to the compressed and original uncompressed data. We adapt a compression method by distances between transmitter and receiver. We show that compressing data for all of the distances is not useful. Experimental results indicate that our approach save power consumption as much as 24%. We also add encoding approach to compressor, to achieve both bit and transition reduction. Experimental results show that by using coder and decoder and adapting part, power saving can reach up to 30%.