Markov Process Based Array Non-Stationarity Modeling for Massive MIMO Channels

For the design and performance evaluation of potential massive multiple input and multiple output (MIMO) related algorithms, accurate channel models are indispensable. Spherical wavefront and array non-stationarity due to physically large antenna array are two new characteristics specific to massive MIMO propagation. In this paper, spherical wavefront effects can be well characterized by a predefined two-dimensional multi-confocal ellipse scattering geometry as in reference [1] and a 9- state Markov process is developed to capture the channel behaviors by array non-stationarity. Meanwhile, the associated approach to derive the state transition probabilities of the Markov process based on the outfield measurements is detailed. Finally, simulation results verify the validity of our proposal.