A Framework for MIMO Capacity Bounds Based on the Cramer-Rao Bound on the Channel Estimation Error

Abstract

This paper provides n framework for deriving MIMO capacity bounds that takes channel estimation error into account by using the Cramer-Rao bound (CRB). Combining the CRB with capacity analysis allows for studies of the trade-off between estimation performance and throughput. The framework is used to investigate the impact of the spatio-temporal structure found in practical channel sounding experiments. Numerical evaluations indicate that higher capacities are possible and the optimum number of training symbols and transmit antennas are different when exploiting the structure in the channel compared to previous unstructured approaches