Performance Evaluation of Scheduling Schemes for D2D Communications

Abstract

We address the performance of a system where D2D communications add up to standard cellular traffic in future 5G networks. D2D resources are taken from the Uplink spectrum and, in this framework, two scheduling schemes are envisaged: the first one consists in two independent schedulers working on separate subsets of the Uplink channel, respectively dedicated to Uplink cellular communications and to D2D communications; the second one is a joint scheduler making use of the whole Uplink channel. In both schemes, all communications are scheduled, which in particular avoids the intra-cell interference issue. We propose multi-class Processor Sharing queueing models which account for realistic cellular radio conditions and provide traffic performance in terms of the average throughput for the different classes of flows. From the considered numerical scenarios, we conclude that the joint Uplink scheduling scheme provides better performance and allows D2D users to make an efficient use of their higher potential capacity, due to the device proximity. The corresponding throughput is also shown to be significantly higher than that perceived without any dedicated D2D link.