Subspace Modeling of Multi-User MIMO Channels

In multiple-input multiple-output (MIMO) systems, the propagation channels can be characterized by their spatial structure, described by the (complex-valued) channel correlation. Accordingly, also interference in MIMO systems is likely to occur in a spatially structured, correlated way. We discuss that structured spatial interference affects mutual information under interference, depending on the eigenspace alignment between the channel of the desired signal and the channel carrying the interference. Intuitively, worst-case interference occurs, when the channels from the intended signal and the interference show a similar spatial structure. In contrast, least hurtful interference is encountered when these channels are maximally non-aligned. In this paper, we develop an analytical channel model generating multi-user MIMO channels with a given degree of severity of interference, described by the alignment of the channels' eigenspaces. Using radio channel measurements, we parameterize our model to reflect realistic scenarios.