A Heuristic Approximation Algorithm to Design Power and Area Efficient FIR Filters in the Canonic Signed Digit Space

This paper introduces a heuristic algorithm that approximates the filter coefficients to the neighbouring values, which have a larger number of zeros in the canonical signed digit (CSD) representation. This approximation increases the number of zeros in the CSD represented filter coefficients so that the realization of the resultant filter requires a reduced number of non-zero partial products in its multipliers, which in turn helps to reduce the area and power consumption of the digital filters. The proposed approach makes sure that the filter performances are maintained within the given limits, and the linear phase property of the filter is preserved. The advantages of the proposed algorithm are demonstrated by designing approximated FIR filters using the algorithm and by analyzing its area and power consumptions and output parameters.