Robust beamforming using a narrowband partially adaptive array

Partial adaptivity speeds convergence and decreases complexity of adaptive interference cancellers, especially designs that use a fixed rank-reducing transformation. The best-performing fixed transformation exploits knowledge of the array response and assumes a given pointing direction. We consider the effects of random complex array errors on the performance of such a fixed transformation. The noise injection and iterative projection methods, used together, make the power space (PS) transformation insensitive to array errors; used singly, performance suffers, and with neither, PS fails unless the weight number matches the interference dimension. Narrowband PS designs are shown to be steerable using conventional steering vectors.