Blind equalizers do not require any known training sequence during the startup period, but can rather perform the equalization based on the received data stream. This project addresses the realtime digital communications problem whereby a digital signal source is transmitting through an unknown continuous noisy propagation medium or channel which is subjected to multipath effects and possibly non-minimum phase. By oversampling the received signal compared to its emission rate, the unknown noisy channel characteristics can be identified and the original signal reconstructed. The technique used to recover the signal is based on the orthogonality property between the signal and noise subspaces. This report presents the results of investigation into the blind approach for the identification of the channel.

Master of Science (Communications and Computer Networking)