Abstract

In this paper, a new technique is presented for the design and optimization of digital FIR filters with coefficients that are presented in canonic signed-digit (CSD) format. Since such implementation requires no multipliers, it reduces the hardware cost and lowers the power consumption. The proposed technique considers three goals, the optimum number of coefficients, the optimum wordlength, and the optimum set of coefficients which satisfies the desirable frequency response and ensures the minimum hardware cost by minimizing the number of nonzero digits in CSD representation of the coefficients using Genetic Algorithms (GA). Comparing with equiripple method, the proposed technique results in about 30-40 percent reduction in hardware cost.