Interference-Aware Downlink and Uplink Resource Allocation in HetNets with D2D Support

We address the resource allocation problem in an LTE-based 2-tier heterogeneous network where in-band D2D communications are supported under network control. The different communication paradigms share the same radio resources, thus they may interfere. We devise a dynamic programming approach to efficiently schedule download and upload traffic, by 1) efficiently matching communicating endpoints and 2) assigning radio resources in an interference-aware manner while accounting for the characteristics of the content to be delivered. To this end, we develop an accurate model of the system and apply approximate dynamic programming to solve it. Our solution allows us to deal with realistic large-scale scenarios. In such scenarios, we compare our approach to today's networks where eICIC techniques and proportional fairness scheduling are implemented. Results highlight that our solution increases the system throughput while greatly reducing energy consumption. We also show that D2D mode, established either in the downlink or uplink, can effectively support delivery of highly popular content without significantly harming macrocell or microcell traffic, leading to increased system capacity. Interestingly, we find that D2D mode can also be a low-cost alternative to microcells.