Power control in D2D-based vehicular communication networks

This paper studies how to efficiently apply deviceto- device (D2D) communications underlaying a cellular system to support vehicle-to-vehicle (V2V) connection (termed D2D-V). By considering the geographic features of the D2D-V system, we propose a D2D-V grouping, reuse channel selection (RS), and power control (PC) framework to achieve the optimal performance of the D2D-V system, in terms of either maximized sum rate or maximized minimally achievable rate. First, in a full channel state information (CSI) scenario, we apply difference of two convex functions (D.C.) programming to obtain the optimal PC. However, because the CSI between some terminals is hard to obtain and some interference can be appropriately inhibited by taking advantage of the vehicles' geographic features, we make a series of simplifications to the PC problemto reduce the requirement of full CSI, the dependence of centralized control, and the computational complexity. The suitable conditions of each simplification are elaborated. Each step of our simplifications are shown as a certain tradeoff between performance and complexity. Furthermore, we provide an investigation into the service quality that one vehicle can achieve when it passes through the covered segment of the highway and propose the location-service curve to portray this. We use numerical simulations to verify the accuracy and feasibility of each step of our suboptimal PC and specific features embedded in the location-service curve.