Integrated Communication and Bayesian Estimation of Fixed Channel States

Abstract

This work studies an information-theoretic performance limit of an integrated sensing and communication (ISAC) system where the goal of sensing is to estimate a random continuous state. Considering the mean-squared error (MSE) for estimation performance metric, the Bayesian Cramér-Rao lower bound (BCRB) is widely used in literature as a proxy of the MSE; however, the BCRB is not generally tight even asymptotically except for restrictive distributions. Instead, we characterize the full tradeoff between information rate and the exact MSE using the asymptotically tight BCRB (ATBCRB) analysis, a recent variant of the BCRB. Our characterization is applicable for general channels as long as the regularity conditions are met, and the proof relies on constant composition codes and ATBCRB analysis with the codes. We also perform a numerical evaluation of the tradeoff in a variance estimation example, which commonly arises in spectrum sensing scenarios.