Reduced complexity blind equalization schemes for ZP-OFDM systems

In this paper, we face the equalization of zero padding (ZP) OFDM transmissions based on blind source separation (BSS). Two new reduced complexity algorithms are presented. These algorithms are based on interpreting the ZP-OFDM system model as a channel-coded cyclic prefix OFDM one. Equalization of this system is equivalent to a BSS problem where the mixing matrix is reduced to the multiplication of a known matrix by an unknown diagonal matrix. We present two alternatives for computing that diagonal matrix using a natural gradient based update equation. These two algorithms have much lower complexity, faster convergence and better performance than other BSS-based equalization algorithms for ZP-OFDM.