Adaptive beamforming in quadratic residue number systems on a cluster array processor

The authors describe a quadratic residue number system (QRNS) implementation of the Leverrier algorithm for computing the adjoint and determinant of a matrix on a prototype parallel signal processor, called the cluster array processor (CAP). Some preliminary performance results for an eight-element antenna array model are also presented. The two-dimensional parallelism of the CAP system provides for truly independent processing of the separate RNS channel, as well as concurrent processing within each channel. For the experiments presented here, it was possible to achieve up to 18 dB of total power reduction, with a processing time of less than 3s