General Optimum Block Adaptative ICA for Signal Separation with Known Source Distributions

T.T. Yang and W.B. Mikhael (USA)


Block adaptation, ICA, Interference Suppression, Time varying conditions


This paper presents a novel implementation of optimum block adaptative Independent Component Analysis (ICA) algorithm for signals with known source distributions, which is frequently a valid assumption in practice. The block algorithm calculates the optimum weight update based on a Taylor Series expansion. The technique imposes no requirement on the number of the sources and block length, and it is suitable for application in time varying environment. The proposed method is simulated for interference rejection in wireless receivers, and the performance is compared with the well-known Fast-ICA algorithm. Simulation results indicate that the new algorithm outperforms Fast-ICA under dynamic conditions at no additional computational cost. For mobile cellular systems, this algorithm lends itself to practical application in the case of high user mobility, handoff and/or rapid changing channel conditions.

Important Links:

Go Back