Application of block-floating point operations in the adaptation of an antenna array

An algorithm for the adaptation of phased arrays is proposed which is based on the inversion of the correlation matrix of samples of noise signals by a completion method using block-floating point arithmetic. The algorithm provides for a high efficiency of noise suppression for a learning-sample volume equivalent to the number of effective noises, with subsequent stabilization of the suppression level with further increase in the learning-sample volume. The hardware costs are comparable to those for fixed-point arithmetic.