On the multi-channel noise cancellation using beamforming algorithm

Abstract

This paper presents performance analysis of beamforming algorithm for canceling multiple channel noise depending on variation of hidden layer of the multilayer feedforward network and the number of epoch (also known as number of iteration). We consider the learning rate variation of the multi layer perception (MLP) to determine the adaptive learning rate of the network. An MLP has been considered to perform beam-forming, which is an array of sensors connected to the MLP inputs. We use the backpropagation algorithm as the learning rule for MLP and improving the signal quality. This involves a desired signal whilst removing any noise or interference signals which may come from different sources.