DoA resolution limits in MIMO channel sounding

Several high-resolution parameter estimation algorithms have been applied to directional channel parameter estimation in channel sounding. However, the variance and reliability of the estimation results are not often clearly defined in practical environments. Especially, imperfections, mutual coupling, and residual calibration errors of real antenna arrays impose limits to the DoA parameter estimation performance. One critical example is the resolution of coherent wave fronts which is often the case in real-time channel sounding. We derive the Cramer-Rao lower bound (CRLB) of an unbiased direction-of-arrival (DoA) estimator. The CRLB indicates the minimum achievable parameter variance which cannot be outperformed by any parameter estimator. It is shown that the required derivatives of the observations with respect to the DoA and path weight parameters can be deduced from a properly defined calibration data model.