A highly parallel architecture for adaptive multichannel algorithms

In this paper an adaptive lattice structured filter for processing wideband spatial data received by an array of sensors is discribed. Using a special architecture that implements a modified version of the Gram-Schmidt orthogonalization procedure a very simple derivation of number of algorithms involving multichannel inputs are presented. A wavefront array processing architecture is described that solve the following algorithms: multichannel spatio-temporal exact least square adaptive lattice filter, matrix-inversion algorithm, linear equations solution and the Fadeev problem. The same architecture and processor elements are shared by all four algorithms.