A blind sequence detection and its application to digital mobile communication

Abstract

A method of sequence detection without the knowledge of the channel response is proposed. The Viterbi algorithm is extended to a blind form by introducing a new branch-metric, defined as the minimum of the short time-average of the squared error (y/spl circ/k-yk)/sup 2/, where yk and y/spl circ/k are the received signal and its replica, respectively. All possible candidate sequences contained in the short time squared error are defined as trellis states, for which the short time squared error is minimized in respect to a variable of the unknown channel response. The proposed implicit blind sequence detection need not keep each variable after the branch metric is calculated. The consistency of the algorithm is justified by proving that a unique sequence is detected in noise free case. The proof is accomplished under condition that the period of the time-average is longer than the channel response. If the additive white Gaussian noise is assumed, the short time squared errors are minimized beyond the desired minimum by the standard Viterbi algorithm using an apriori known channel response. We call this phenomenon over-minimization. The over-minimization is a major reason of the unavoidable error rate degradation in the blind receiver. An objective of this paper is to establish blind sequence detection for digital mobile communication. Provided the channel response can be regarded as time-invariant during the period of the short time-average of the squared error, the blind sequence detection keeps the same performance as in time-invariant case. In order to improve the error rate performance, an algorithm based on the fractional sampling scheme is introduced. Several error rate performances and behaviors of error events are investigated by computer simulation.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>