Analysis and Modelling of Massive MIMO Mobility Channels

Abstract

The demand for capacity within existing mobile networks continues to increase as more subscribers and more devices communicate and as data-rich applications become more popular. The evolving 5G telecommunications standards aim to respond to such demand. A promising approach to increasing capacity and reliability within the context of 5G is Massive Multiple-Input Multiple-Output (MIMO) where many transmit antennas are used relative to the number of users, thus providing a greater opportunity to use the spatial characteristics of the channel for spatial diversity and multiplexing. The performance and modelling of Massive MIMO scenarios with mobile users is of great importance and this paper presents results from both simulated and practical mobility campaigns with multiple-users and large base station antenna arrays. The evolution of the Condition Number in time for the simulated and practical scenarios is examined as a way of quantifying the multi-path richness of the channel and the rank deficiency of the channel correlation matrix, an important feature as a full rank correlation matrix represents the ideal scenario for Massive MIMO based spatial multiplexing, within a mobility scenario. A phenomenon is identified in both measured and simulated data in which the Massive MIMO channel appears to pass through several distinct stages where the features of the channel change significantly. A Change Detection algorithm is presented as a way of identifying and formalising the presence of these changes, an analysis which leads to a demonstration of the use of Markov Chains for the modelling of Massive MIMO mobility channels. Further analysis is presented to determine the feasibility of the use of an Autoregressive Moving-Average process as a possible alternative to the use of Markov Chains.