Efficient Data Processing in Software-Defined UAV-Assisted Vehicular Networks: A Sequential Game Approach

In large scale networks like Vehicular Ad-hoc Networks (VANETs), the full coverage of fixed infrastructure is hard to ensure, making network management difficult. Whether in infrastructure-less environments where the network connectivity is poor or where the infrastructure deployment is difficult, costly or not profitable. Recently, in the one side, Unmanned Aerial Vehicles (UAVs) have been used as a new flexible solution to assist infrastructure-less vehicular networks for the investigation of inaccessible areas. In the other side, several works have shown interest in the use of the emerging network paradigm of Software-Defined Networking (SDN) to facilitate the management and improve the performances of vehicular networks. In this paper, we propose a novel distributed SDN-based architecture for UAV-assisted infrastructure-less vehicular networks. The main goal is to fill the gap that no SDN-based architecture has been proposed for these networks. We focus particularly on a road safety use-case that incorporates UAVs to assist emergency vehicles in the exploration of affected zones in critical emergency situations. Moreover, we investigate how to achieve efficient data processing policy through a computation offloading/sharing decision-making problem. The main challenge is to reach the best tradeoff between computation delay and energy consumption for computation-intensive tasks in a delay-sensitive context. We formulate this decision problem as a two-player sequential game approach and design distributed computation algorithms to solve the problem. Numerical results show that data processing policy of distributed offloading/sharing algorithms achieves efficient computation performances in terms of delay and energy whilst ensuring until 28% gain of system cost and 95% better response time, compared to native computation scenarios and related data delivery UAV-assisted VANET works, respectively.