ABSTRACT

Abstract. In this work, we present and analyze a number theoretic approach to computing one-dimensional cyclic convolution of sequences defined in finite integer and complex integer rings. A fundamental result of this work is that under the non-restrictive condition, (N, M) = 1, the algorithms defined in finite integer and complex integer rings are as intensive computationally as the corresponding algorithms defined in rational and complex rational number system only in the worst case. They simplify considerably for a large number of cases of importance in digital signal processing.