Asymptotically Efficient Multichannel Estimation for Opportunistic Spectrum Access

The problem of estimating the parameters of multiple independent continuous-time Markov on-off processes is considered. The objective is to minimize the total mean square error (MSE) under a constraint on the total sensing time. The Fisher information matrix for the primary traffic model and the maximum likelihood estimator are obtained. A sequential estimation strategy is proposed which operates under an epoch structure with growing epoch length. Specifically, the total sensing time in the current epoch is allocated among the on-off processes based on the current estimates of the parameters using observations obtained in previous epochs. It is shown that this sequential estimation strategy is asymptotically efficient as the total sensing time increases. This result finds application in opportunistic spectrum access where secondary users need to estimate the channel occupancy model of the primary system for efficient exploitation of spectrum opportunities.