Blind source separation for OFDM with filtering colored noise out

摘要

Two blind algorithms that are developed with the intention of improving the symbol detection of Orthogonal Frequency Division Multiplexing (OFDM) techniques are proposed in this paper. OFDM systems are easy to equalize in implementations. The schemes are based on the theories of blind source separation (BSS). They are among the premier mechanisms used for extracting unobserved signals from observed mixtures in signal processing. In this study noise component of the received signal mixture is tried to be filtered out. A scalar energy function with the iterative fixed point rule for receive signal is used in determining the filter coefficients while taking the time correlation properties of the channel to advantage for BSS. The methods are tested in a single input single output slow fading channel with a receiver containing Equal Gain Combining (EGC) for treating the channel state information (CSI) values. These solutions can be introduced as low computational complexity approaches.