Reducing complexity of fixed-coefficient FIR filters

An approach for reducing the complexity of the multiplier block for fixed coefficient FIR filters using the transpose structure is presented. Through the use of modular arithmetic with moduli of the type 2ⁿ±1 and 2ⁿ, modulo multiplier blocks with a small full adder count are obtained using a minimum spanning tree algorithm. The results obtained show a very significant reduction in the filter complexity relative to existing techniques. A unique feature of the results is that the complexity is independent of filter length, but a function of the filter's dynamic range.