Device-to-Device (D2D) Communication: Fundamentals with Applications to LTE

09 Jun 2014

Device-to-device (D2D) communication that enables direct communication between nearby mobiles is an exciting and innovative feature of next-generation cellular networks. It will facilitate the interoperability between critical public safety networks and ubiquitous commercial networks based on e.g. LTE. How should we analyze and design such hybrid networks consisting of both cellular and ad hoc links? WNCG Profs. Jeffrey Andrews and Constantine Caramanis, in collaboration with lead researchers and WNCG students Xingqin Lin and  Qiaoyang Ye are working to develop novel models for such hybrid cellular systems and to use them for network performance evaluation and design.  This research involved close collaboration with and significant technical inputs from the Head of NSN North America Radio Systems, Amitava Ghosh, and Principal Engineer Mazin Al-Shalash from WNCG Industrial Affiliate Huawei. Their work has impacted 3GPP standards contributions from NSN and Huawei on specific D2D technologies in LTE Release 12.

In principle, exploiting direct communication between nearby mobile devices will improve spectrum utilization, overall throughput, and energy efficiency, while enabling new peer-to-peer and location-based applications and services. D2D-enabled LTE devices have the potential to become competitive for fallback public safety networks that must function when cellular networks are not available or fail. Introducing D2D poses many new challenges and risks to the long-standing cellular architecture, which is centered around the base station (BS). In a recent magazine paper, Prof. Jeff Andrews, Xingqin Lin, Amitava Ghosh, and Rapeepat Ratasuk provided an overview of D2D standardization activities in 3GPP, identified outstanding technical challenges, drew lessons from initial evaluation studies, and summarized “best practices” in the design of a D2D-enabled air interface for LTE-based cellular networks.

Among others, one fundamental issue is how to share spectrum resources between cellular and D2D communications. Specifically, should D2D mobiles use orthogonal spectrum resources or opportunistically access the spectrum resources occupied by cellular mobiles? Spectrum sharing is further complicated by the new design flexibility of D2D mode selection that means a potential D2D pair can switch between direct and conventional cellular communications. In two recent papers, Prof. Jeff Andrews, Prof. Constantine Caramanis, Xingqin Lin, Qiaoyang Ye, and Amitava Ghosh, Mazin Al-Shalash used tools from stochastic geometry to address the D2D spectrum sharing problem, proposing novel tractable models for hybrid networks, and developing a unified analytical framework for performance analysis and design of hybrid networks. They successively applied the proposed models and the analytical framework to the study of the diverse D2D spectrum sharing scenarios.

The first paper considers sharing uplink spectrum with D2D mobiles and derives analytical rate expressions. The WNCG team found that D2D mobiles enjoy much higher data rates than regular cellular mobiles due to the short range of communications. Cellular mobiles may also benefit from D2D as D2D can help offload traffic from congested cellular networks. From a coverage perspective, we revealed an interesting tradeoff between D2D spectrum access and mode selection: as more potential D2D mobiles use direct communication mode, the network should actually make less spectrum available to them to limit their interference.

As a parallel work to the first paper, the research team investigated a D2D-enabled cellular network, where downlink resources are either partitioned or shared between D2D and downlink cellular transmissions. They provided tractable and accurate analytical results that are amenable to efficient optimization. The researchers found that to maximize the total throughput, D2D links with more traffic to transmit should be more aggressive in their spectrum access, despite the interference this generates to the rest of the network. In a heavily loaded network, the total throughput benefits from offloading local traffic to D2D mode, as D2D communication only requires “1” hop while relaying via a BS requires “2” hops. They further investigated the optimal resource partition between D2D and cellular networks, and found that the choice of dedicated and shared approaches depends on the D2D traffic and the resource partition in dedicated networks. The dedicated approach may achieve larger throughput in a network with many short-range D2D links and optimal resource partition.

This research supported in part by the National Science Foundation.