Adaptive beamforming in the presence of correlated arrivals

The constrained least-mean-square (LMS) algorithm updates the weights of an antenna array by estimating the gradient of the mean output power surface. In its standard form an estimate of the required gradient is made by multiplying the array output with the sensor outputs. This form of LMS algorithm is not suitable for updating the weights in the presence of correlated arrivals. The correlation between the desired signal and an unwanted interference may exist in situations of multipath and deliberate jamming. This paper investigates the performance of two algorithms, namely, the structured gradient algorithm and the improved LMS algorithm, in the presence of correlated narrow-band arrivals. Theoretical analysis and computer simulation are presented to demonstrate the capability of these algorithms to estimate the weights of a linear array of equispaced elements, to cancel the correlated directional sources.