Efficient hardware architectures for eigenvector and signal subspace estimation

We consider hardware solutions to the adaptive-signal subspace-estimation problem. In deriving a hardware-realizable subspace tracking algorithm, we have applied delayed updating to the PASTd algorithm to achieve high speed. Pipelined and systolic architectures and the estimation of the dominant eigenvector or the signal subspace are also studied. Methods for approximating a reciprocal computation are employed and simulation results are presented to validate our algorithm and hardware architectures.