A fast adaptive algorithm based on the gradient method for the reduction of the computational requirements

The performance of the adaptive algorithm is mainly evaluated by the following: 1) the characteristics for the convergence and 2) the reduction of the computational requirements per sample time. The algorithm is composed by the block signal processing and the orthogonal projection which satisfy these requirements. Furthermore, a method to improve convergence speed of the algorithm is given; the "shifting" technique. And, we can consider the adaptive algorithm which can be derived by introducing the "shifting" technique. This algorithm has faster convergence speed. The purpose of this paper is to present a new adaptive algorithm whose computational requirements are less than that of the original algorithm.