Joint data detection and channel estimation for rapidly fading channels

Two algorithms for joint data detection and channel estimation are proposed. These algorithms can be implemented in communication schemes where typically two-to-four level signaling is used and the channel exhibits very rapid changes in its characteristics. Both algorithms converge to the channel estimate within 100 data symbols for binary pulse-amplitude modulation (PAM) with SNR down to 10 dB. One employs blind trellis search; maximum-likelihood sequence estimation is done without the channel impulse response being known initially, but the estimate is gradually updated during the blind trellis search. The other is based on maximization, of the average likelihood function. The solution is obtained by an iterative technique based on the Baum-Welch algorithm. The performance of these algorithms is evaluated for 2-PAM signaling over three typical channel types with various noise levels