A method for nonasymptotic performance analyses of array signal processing algorithms

This work features a stochastic perturbation theoretic approach that can be used to calculate the performance analyses of array signal processing algorithms for the nonasymptotically large, moderate data regimes that are appropriate for many practical system applications. As an example, the perturbation method is used to develop the expressions for the moments of the MUSIC null spectrum at moderate SNR for independent data snapshot numbers as low as /spl sim/5. Theoretical predictions are shown to be in good agreement with empirical statistical simulation results.<
>