Low-energy software Reed-Solomon codecs using specialized finite field datapath and division-free Berlekamp-Massey algorithm

This paper presents hardware/software codesign of low-energy programmable Reed-Solomon (RS) codecs. These universal RS(n, k) codecs are to be implemented as a combination of hardware and software in application-specific DSP processors with specially designed programmable finite field datapath and dedicated and optimized software to reduce the total energy consumption. A novel frequency-domain RS decoding procedure using division-free Berlekamp-Massey algorithm is presented. Universal RS decoders are coded using this decoding algorithm as well as other frequency-domain and time-domain decoding schemes based on three types of datapath architectures. From extensive experimental results and cross-comparisons of both energy and energy-latency products, it is concluded that RS decoders using the proposed frequency-domain RS decoding procedure with division-free Berlekamp-Massey algorithm based on finite field datapath with separate MAC and DEGRED units have the best performance.