A novel transformation for reduction of switching activity in FIR filters implementation

A novel high level transformation for low power implementation of FIR filters is presented. The new idea is the reordering of the filter coefficients aiming at the minimisation of the switching activity. As a measure of the switching activity the Hamming distance (HD) between successive coefficients, stored in a memory, is used. The transformation can be incorporated both in application specific architectures and in general purpose programmable architectures. The reordering of the N coefficients, for the HD optimisation of their sequence, can be modelled by a travelling salesman problem (TSP), which is a well known NP-complete problem. A novel heuristic algorithm for a fast and accurate solution to this problem is proposed. Experimental results show that the proposed technique leads to significant power savings in terms of switching activity reduction