Using system-level simulation to evaluate downlink throughput performance in LTE-A networks with clustered user deployments

Downlink throughput performance in Long Term Evolution - Advanced (LTE-A) networks has been widely investigated by many simulation studies. However, these studies generally assume that users are uniformly distributed in regularly arranged hexagonal cells. Analytical models and methods of stochastic geometry have been applied to wireless networks to consider the effects of spatial configuration of nodes and base stations on network performance by abstracting away many details of the systems. As such, comprehensive studies of the effects of spatial factors on throughput performance using standard-compliant, open-source software that simulates many details of LTE-A networks are still scarce. In this paper, we use the Vienna LTE-A Downlink System Level Simulator to re-evaluate the validity of usual throughput performance metrics under various spatial deployment scenarios and employ statistical methods to determine the significance of performance observations. Our primary aim is to show that clustered user deployments around a base station exhibit significantly different throughput performance compared to the case of uniformly distributed users throughout the cell. This is important since in practice, base stations are deployed in areas of high user density. In addition, we show that the most widely used scheduling scheme in cellular networks, namely the proportional fair scheduling, may lose its performance advantage over round robin scheduling when users are clustered around the base station. These findings are expected to have implications on the design and deployment of next generation cellular networks.