A comparison of blind channel estimation and equalisation techniques for a fading environment

A number of blind channel estimation techniques has previously been presented. We consider the performance of blind techniques for a time-varying, fading, environment. The three techniques we investigate are Tong´s (1994), subspace, and multiple survivor Viterbi methods due to their fast convergence and robustness to additive Gaussian noise. Their performance is compared with the non-blind cross-correlation scheme used in the GSM system. We also give the bit error rates (BER) in terms of the SNR after passing the resulting channel estimates from the three techniques to the Viterbi decoder. It is shown that both Tong´s and the subspace methods outperform the multiple survivor Viterbi algorithm. Also, Tong´s and subspace method show only a 5 dB loss in SNR as compared with the non-blind cross-correlation technique