Performance analysis of adaptive arrays with projected perturbation sequences

Abstract

Perturbation techniques are useful in the design of low complexity adaptive antenna arrays for estimating the gradient required in stochastic descent algorithms. Implementing projected perturbation sequences in an adaptive array allows the simultaneous reception of signals and the adaptation of the array weights while preserving the constraints imposed on the array weights.This thesis quantifies the performance of narrowband adaptive array processors that employ projected perturbation techniques. For different perturbation receiver structures the performance is determined under idealised conditions and importantly also when practical implementation issues are taken into account.The arrays performance is characterised by analysing the transient performance of the weight covariance matrix and by determining the misadjustment. By drawing similarities between two established analysis techniques a new misadjustment analysis technique is introduced.Practical implementation can impact on the arrays performance such that the benefit of the projected perturbation approach is lost. By characterising the array's sensitivity to perturbation noise additional projections which counteract some implementation effects are identified. The level of loss of performance due to weight quantisation and the limited dynamic range of the array weights is determined.