Stochastic analysis of cooperative mimo techniques for downlink interference mitigation in wireless cellular networks
Kianoush Hosseini · TSpace (University of Toronto) · 2015
Interference is one of the key factors limiting the performance of wireless cellular networks with dense base-station (BS) deployments. This thesis studies two prominent techniques for interference mitigation and improving data rates of cellular networks referred to as: (1) network multiple-input multiple-output (MIMO) and (2) large-scale MIMO (LS-MIMO). The primary objective is to analyze the benefits and limitations of these two schemes in the downlink of a cellular system that comprises multiple-antenna BSs and single-antenna users, adopts zero-forcing beamforming, and employs equal power allocation. There are three parts to this thesis. The first part develops an analytical tool to study the cooperation gain of network MIMO systems with disjoint clusters. It is shown that network MIMO has two major limitations. First, due to the need to perform zero-forcing from a distributed set of BSs, a network MIMO system introduces a penalty in terms of the received signal power. Further, users located closer to the cluster boundaries are prone to significant interference, which considerably constrains system performance. The next two parts of this thesis propose solutions for these limitations. The second part shows that the signal power penalty can be mitigated if the LS-MIMO technique is adopted for interference mitigation. In particular, we prove that an LS-MIMO system is superior to a comparable network MIMO system for a wide range of utility functions. In the final part of this thesis, an alternative BS clustering termed as user-centric clustering is considered. Under this clustering strategy, BS clusters are formed for each user independently. Consequently, cluster boundaries no longer exist. We quantify the performance of an LS-MIMO system with user-centric clustering, and explore different avenues in which such a clustering scheme brings significant performance gains. Further, the tradeoff among providing sum-rate multiplexing, diversity, and interference cancellation in such a system is studied. It is shown that a close-to-optimal strategy for maximizing the per-BS ergodic sum rate is to use none of the spatial resources for interference cancellation, while selecting the number of users to schedule and diversity order per user properly.