Sequence estimation over linearly-constrained random channels.
This paper presents a new approach using EM(Expectation-Maximization) algorithms for ML (maximum like-lihood) sequence estimation over unknown ISI (inter-symbol in-terference) channels with linearly-constrained random channelcoefficients which may be fast time-varying. By using the EMformulation to marginalize over the underlying channel coeffi-cient distribution, maximum-likelihood estimates of the transmit-ted sequence are obtained. The EM algorithms are shown to per-form better, in terms of BER, than existing algorithms which per-form jointly-optimal sequence and channel estimation, or whichdo not take into account fast time-varying channel effects.
|Main Author:||Perry, Richard.|
|Other Authors:||Buckley, Kevin.|