Sparse multipath MIMO channels: Performance implications based on measurement data

The main focus of the present contribution is on the statistical assessment of Multiple-Input Multiple-Output (MIMO) systems from the viewpoint of underlying multipath sparsity. In contrast to the implicit assumption of rich multipath in the prevalent i.i.d. MIMO fading model, recent investigations and measurement campaigns have shown that physical channels exhibit a sparse multipath structure that in turn leads to spatial correlation. Using a set of measured data and a well-known analytical model, the measured MIMO channel matrix is decomposed in terms of fixed virtual transmit and receive angles, thereby providing a simple geometric interpretation of the scattering environment. It is demonstrated that the presence of direct or obstructed Line-of-Sight (LoS) components strongly correlates with channel sparsity; consequently, the number of spatial degrees of freedom (DoF) is severely limited compared to that in the i.i.d. model. In the sequel, we investigate the relationship between the channel sparsity structure and the mean of propagation paths and quantify the impact of channel sparsity on two key MIMO channel metrics: ergodic capacity and diversity level. In all cases, the analysis of measured data reveals an approximate concave quadratic variation in capacity, diversity and mean number of paths as a function of the number of DoF, thereby suggesting a fundamental inter-dependency between all these features.