Adaptive beamforming with instantaneous convergence & its efficient implementation in DSP hardware

In applications where the scenario is highly nonstationary, lower convergence time of the adaptive algorithm becomes a crucial requirement. In this paper, an adaptive beamformer with a near instantaneous convergence is discussed. Thus more frequent updating of the beamformer weight vector is possible to cater for the dynamic scenario. But this necessitates a much higher computational requirement also. The computational complexity is mainly contributed by the matrix inverse computation. The performance of different algorithms, including the one using Householder transform based Cholesky factorization is compared. A computationally efficient implementation and its mapping to the TigerSHARC Digital Signal Processor (DSP) are presented.