Massive MIMO Channel Modeling

Massive MIMO has attracted many researchers’ attention as it is a promising technology for future 5G communication systems. To characterize the propagation channels of the massive MIMO and to evaluate system performance, it is important to develop an accurate channel model for it. In this thesis, two correlative models, i.e., the Kronecker model and the Weichselberger model, and a cluster-based model, i.e., the Random Cluster Model (RCM), have been validated based on real-life data from four measurement campaigns. These measurements were performed at Lund University using two types of base station (BS) antenna arrays, a practical and compact uniform cylindrical array (UCA) and a physically-large virtual uniform linear array (ULA), both at 2.6 GHz. For correlative models, performance metrics such as channel capacity, sum-rate and singular value spread are examined to validate the model. The random cluster model, which is constructed and evaluated on a cluster level, has been parameterized and validated using the measured channel data. The correlative models are relatively simple and are suitable for analytical study. Validation results show that correlative models can reflect massive MIMO channel capacity and singular value spread, when the compact UCA is used at the base station and when users are closely located. However, for the physically-large ULA, correlative models tend to underestimate channel capacity. The RCM is relatively complex and is usually used for simulation purpose. Validation results show that the RCM is a promising model for massive MIMO channels, however, improvements are needed.