Tracking symbol-valued output sequences

Given a discrete-time state-space linear system whose available outputs are instantaneous discrete-valued functions of the system´s state, the author considers the problem of choosing a feedback control law so as to make the record of discrete state measurements track a given sequence of symbols from the set in which these measurements take values. The problem of optimizing different measures of tracking fidelity for symbol sequences is considered. The control strategies used are of two principal types: the first, traditional schemes based on linear reasoning that regard the discrete measurement history as generating a sequence of approximations to the system´s state, and second, strategies that regard the discrete measurement record simply as a sequence of partial observations whose utility can be enhanced by manipulation of the control