A sparse RNS system

The residue number system, or RNS, has attracted increased attention recently. Using the concurrent processing capabilities of the RNS, high-speed digital filters have been designed. However, the RNS uses memory table lookup operations to perform its arithmetic. Therefore, the RNS can be a prodigious consumer of semiconductor memory chips. In this paper, a procedure is presented which will admit the design of an RNS filter with a reduced memory budget. This concept is called the sparse RNS.