Semi-Blind Channel Estimation for MIMO Diffusive Molecular Communication

Multiple-input multiple-output (MIMO) molecular communication (MC) systems can provide significantly better data rates than single-input single-output (SISO) systems. A major challenge in MIMO MC systems is the presence of inter-link interference, in addition to inter-symbol interference. Accurate estimation of all the channel information is thus essential for interference mitigation in MIMO MC systems. The pilot-based estimators proposed in existing works impose a heavy training overhead. In contrast, we address this problem using semi-blind estimation strategies. In particular, we propose one semi-blind estimator based on the expectation-maximization (EM) criterion. Moreover, as alternatives to the iterative EM estimator, we develop two other estimators that are based on the decision-directed criterion. As a benchmark on estimation accuracy, we derive the semi-blind Cramer-Rao bound. Through simulation studies, we demonstrate that our proposed estimators outperform the previously proposed pilot-based estimators. Moreover, they can be used to considerably reduce the number of transmitted pilot symbols, while maintaining high estimation accuracy.