Title :
A high-performance architecture for an RLS-like eigenvector algorithm
Author :
Xu, Fan ; Willson, Alan N., Jr.
Author_Institution :
California Univ., Los Angeles, CA, USA
Abstract :
We present a high-performance realization of the PAST algorithm for tracking the largest eigenvector of a correlation matrix. Delayed updating is applied to generate the necessary latches for pipelining. A step size approximation scheme is adopted to replace the division operation normally required by the algorithm. The resulting architecture is simulated and synthesized to assess its validity and efficiency
Keywords :
adaptive signal processing; eigenvalues and eigenfunctions; identification; least squares approximations; pipeline processing; PAST algorithm; RLS-like eigenvector algorithm; adaptive filter; data correlation matrix; delayed updating; efficiency; high-performance architecture; largest eigenvector tracking; latches; pipelining; signal subspace identification; signal subspace tracking; step size approximation; Adaptive filters; Convergence; Cost function; Delay; Eigenvalues and eigenfunctions; Finite impulse response filter; Hardware; Pipeline processing; Resonance light scattering; Spectral analysis;
Conference_Titel :
Signal Processing Proceedings, 2000. WCCC-ICSP 2000. 5th International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-5747-7
DOI :
10.1109/ICOSP.2000.894553
