Blind channel identification with modulation induced cyclostationarity

Previous results have pointed out the importance of inducing cyclostationarity at the transmitter for blind identification and equalization of communication channels. The present paper shows that by modulating the input sequence with a deterministic and periodic sequence, single-input single-output channels can be identified uniquely from cyclic second-order output statistics, irrespective of the location of channel zeros, color of additive stationary noise, or channel order over-estimation errors, provided that the period of modulation induced cyclostationarity is greater than half the channel length. Linear, closed-form, linear inverse, and nonlinear correlation matching approaches are developed for channel estimation and are tested using simulations