Stochastic Modeling and Analysis of User-Centric Network MIMO Systems

TL;DR

This paper analytically characterizes the performance of user-centric network MIMO systems, showing that cooperation improves signal quality and reduces interference, with significant benefits for downlink edge users.

Contribution

It introduces an analytical framework for user-centric network MIMO, quantifying the effects of clustering and cooperation on system performance.

Findings

BS cooperation significantly outperforms single-cell processing.

User-centric clustering mainly benefits DL edge users.

Analytical results assume perfect CSI and infinite backhaul.

Abstract

This paper provides an analytical performance characterization of both uplink (UL) and downlink (DL) user-centric network multiple-input multiple-output (MIMO) systems, where a cooperating BS cluster is formed for each user individually and the clusters for different users may overlap. In this model, cooperating BSs (each equipped with multiple antennas) jointly perform zero-forcing beamforming to the set of single-antenna users associated with them. As compared to a baseline network MIMO systems with disjoint BS clusters, the effect of user-centric clustering is that it improves signal strength in both UL and DL, while reducing cluster-edge interference in DL. This paper quantifies these effects by assuming that BSs and users form Poisson point processes and by further approximating both the signal and interference powers using Gamma distributions of appropriate parameters. We show that BS cooperation provides significant gain as compared to single-cell processing for both UL and DL, but the advantage of user-centric clustering over the baseline disjoint clustering system is significant for the DL cluster-edge users only. Although the analytic results are derived with the assumption of perfect channel state information and infinite backhaul between the cooperating BSs, they nevertheless provide architectural insight into the design of future cooperative cellular networks.