A design method of parallel fast RLS second-order adaptive Volterra filter

The adaptive Volterra filter (AVE) is attractive in adaptive filtering applications because its expansion is a linear combination of the input and output signals. However, the formidable computational work of AVE is prohibitive for practical applications. In this letter, we present a parallel fast recursive least squares (RLS) second-order adaptive Volterra filter (PAVE) to reduce computational load. Our discussion is based on the approach of the fast RLS AVE [3], by which the computational complexity has been reduced to O(N-3) multiplications per time instant, where O((.)) denotes '' order of,'' and N is the filter length. Proposed PAVE consists of several subfilters partitioned from the conventional AVE with parallel implementation, the computational work can be reduced effectively. Several simulation results are presented to validate the proposed method.