Study on Scheduling Techniques for Ultra Dense Small Cell Networks

TL;DR

This paper evaluates scheduling techniques in ultra dense small cell networks, showing that simple round robin scheduling may be preferable due to reduced multi-user diversity and increased interference at low inter-site distances.

Contribution

It models Rician fading with distance-dependent K factor and introduces a shadowing model considering inter-site distance, providing insights into scheduler performance in dense deployments.

Findings

Multi-user diversity gain is limited at low ISDs.

Round robin scheduling performs nearly as well as proportional fair in dense networks.

Network densification still improves capacity despite increased interference.

Abstract

The most promising approach to enhance network capacity for the next generation of wireless cellular networks (5G) is densification, which benefits from the extensive spatial reuse of the spectrum and the reduced distance between transmitters and receivers. In this paper, we examine the performance of different schedulers in ultra dense small cell deployments. Due to the stronger line of sight (LOS) at low inter-site distances (ISDs), we discuss that the Rician fading channel model is more suitable to study network performance than the Rayleigh one, and model the Rician K factor as a function of distance between the user equipment (UE) and its serving base station (BS). We also construct a cross-correlation shadowing model that takes into account the ISD, and finally investigate potential multi-user diversity gains in ultra dense small cell deployments by comparing the performances of proportional fair (PF) and round robin (RR) schedulers. Our study shows that as network becomes denser, the LOS component starts to dominate the path loss model which significantly increases the interference. Simulation results also show that multi-user diversity is considerably reduced at low ISDs, and thus the PF scheduling gain over the RR one is small, around 10% in terms of cell throughput. As a result, the RR scheduling may be preferred for dense small cell deployments due to its simplicity. Despite both the interference aggravation as well as the multi-user diversity loss, network densification is still worth it from a capacity view point.