Engineering a Game Theoretic Access for Urban Vehicular Networks.
Zhou Su,Yilong Hui,Tom H. Luan,Song Guo
DOI: https://doi.org/10.1109/tvt.2017.2700049
IF: 6.8
2017-01-01
IEEE Transactions on Vehicular Technology
Abstract:Drive-thru Internet has emerged as a fundamental approach for information/content distribution to vehicles on the go. In the drive-thru Internet, one vehicle may connect to multiple roadside units (RSUs) along its trip, and in the meantime, multiple vehicles may compete one RSU at the same time for transmissions. Therefore, both vehicles and RSUs need to optimize their connections to achieve their best utilities. In the perspective of RSUs, it is important to select optimal vehicles to transmit so as to maximize the global RSU system's utilization. For individual vehicles, it needs to wisely select RSUs along its trip to connect so as to minimize its download cost, e.g., energy consumption and bandwidth cost. This paper targets to address the two design goals in one framework using a game theoretic model. Specifically, we model the two-dimensional drive-thru Internet as a second-price sealed-bid auction. In each RSU cell, an adaptive reserve price scheme is designed such that the RSU is allowed to selectively provide connections to vehicles based on the network size, vehicle's transmission rate, urgency of download, and content popularity; the RSU finally obtains the optimal utility on a Bayesian Nash equilibrium of the auction. For individual vehicles, a finite-horizon Markov decision process has been developed to guide vehicles to optimally select RSUs to connect along their road trips. Using extensive simulations, we demonstrate that the proposed framework can achieve the highest utility for the RSUs compared with existing proposals. It can also help vehicles keep a higher utility and transmission ratio when going through a single RSU or multiple RSUs than the conventional schemes.