Search for a command to run...
A new adaptation algorithm designed for real-time data processing in large antenna arrays is presented. The algorithm is used to determine the set of filter coefficients (weights) which minimizes the mean-square error in a multidimensional linear filter. The algorithm forms an estimate of the target signal, which is assumed to be of interest, in the presence of interfering noises. It is assumed that the direction of arrival and spectral density of the target signal are known a priori. No such information is assumed to be available regarding the structure of the interfering noise field. The a priori target information is incorporated directly into the adaptation procedure using a modified gradient descent technique. The mathematical convergence properties of the algorithm are presented and a computer simulation experiment is used as an illustration. It is shown that as the number of iterations becomes large, the expected value of the adaptive solution converges to the minimum mean-square-error solution. It is further shown that the variance of the adapted filter about the optimum solution can be made arbitrarily small by appropriate choice of a scalar constant in the algorithm. These results are based on the assumption that the array signals are Gaussian and that successive time samples are statistically uncorrelated. Thus, the new algorithm is shown to converge to the optimum processor in the limit as the number of adaptations becomes large. Any disadvantage which may arise in the use of such an asymptotically optimum system is offset by the extreme simplicity of the adaptive procedure. This simplicity should prove to be particularly useful in many of the practical array processing problems recently encountered in seismic and sonar data processing.