Optical Processing Of Covariance Matrices For Adaptive Processors

This paper details the use of linear algebraic optical processors for narrowband and wideband adaptive phased array radar (APAR) applications. A general APAR scenario is explained, outlining the need for improvements in processor performance which optical processing techniques can provide. We then analyze the optical architecture and its implementation including negative base encoding for bipolar data, AC-coupled inputs for linearity and temperature stability, matrix partitioning for handling large matrices, and bit partitioning for improved accuracy. A new block Toeplitz processing algorithm for wideband processing is presented. Results of an inverse covariance matrix updating algorithm are shown.