An adaptive system for direct blind multi-channel equalization

Blind equalization has important applications to wireless communication and cable HDTV. It is known that the output of the channel bank has sufficient information for blind identification, and hence also for blind equalization It may be surprising to find that the composite output of the equalizer bank indeed has sufficient information for blind equalization, though it might not have sufficient for blind identification. Specially, if the zero-mean transmitted signal is temporally uncorrelated and if the channel bank has no common nonzero zeros, the equalizer bank equalizes the channel bank if, and only if, its composite output is temporally uncorrelated. An adaptive algorithm is introduced in this paper as an implementation of above theorem, and computer simulations demonstrate its fast convergence (less than 100 iterations) and good tracking ability (for simulated 70 mph mobile channels with SNR=15 dB).