OPTIMUM BLOCK ADAPTIVE ICA FOR SEPARATION OF REAL AND COMPLEX SIGNALS WITH KNOWN SOURCE DISTRIBUTIONS IN DYNAMIC FLAT FADING ENVIRONMENTS

Efficient co-channel and adjacent channel interference rejection is often one of the most demanding requirements for wireless receivers. Independent Component Analysis (ICA) has been previously applied to realize interference suppression. In particular, the fixed-point FastICA and complex FastICA algorithms can successfully perform blind signal extraction for real and complex valued communication signals in stationary or slow fading environments. Both algorithms exhibit fast convergence speed and impressive accuracy due to their Newton type fixed-point iteration. However, under dynamic channel conditions often encountered in practice, the fixed-point algorithms' performance is significantly degraded. In this contribution, a novel complex block adaptive ICA algorithm and its simplified real version is proposed, that overcome this limitation for the separation of complex valued and real signals with known source distributions. The new methods exploit prior information about the modulation scheme of the communication signals of interest, and achieve improved interference suppression performance in dynamic channel environments. The proposed complex ICA algorithm is called Complex Optimum Block Adaptive ICA (Complex OBA-ICA), and its abridged version for separating real signals is called General Optimum Block Adaptation ICA (GOBA-ICA). The proposed methods are applied to interference rejection in linearly and abruptly flat fading dynamic environments for diversity QPSK and BPSK wireless receivers. Simulation results show that the presented techniques demonstrate better convergence properties and accuracy as compared to the complex FastICA and FastICA algorithms.