Stable Noise Subspace Estimation Algorithm Suitable for VLSI Implementation

Abstract

The aim of this paper is twofold: 1) to stabilize a fast adaptive noise subspace algorithm, which was proposed earlier in the literature under the name of FRANS algorithm; and 2) to present a low computational cost version that is free from any square root and division operations, which is very suitable for VLSI implementation. A theoretical mean-square analysis and simulation results are also given in order to assess the performance of this algorithm.