Deployment Optimization of Data Centers in Vehicular Networks

Abstract

Due to the ubiquitous utilization of GPS devices, traffic cameras, and sensing devices, data are collected more readily in a smart city. If all the cabs of this city are used as data carriers, data generated by sensing devices will be collected to data centers efficiently and economically. Therefore, vehicular networks joint sensing devices provide more perspectives for a variety of vehicle-based applications. In this paper, three data centers deployment optimization schemes are proposed to optimize data centers' deployment, which can considerably enhance the performance of data collection and code dissemination tasks. Each proposed scheme uses different criteria to optimize the deployment of data centers: 1) locations of sensing devices; 2) locations that have high traffic flow; and 3) locations that have high valid flow, and these schemes are called Scheme 1, Scheme 2, and Scheme 3, respectively. In addition, another scheme that produces the deployment of data centers randomly is called Scheme 4, and it is used as a contrast in experiments. After performing extensive experiments and simulations based on two real-world datasets of cabs' GPS coordinates, the experiment results demonstrate that Scheme 3 noticeably outperformed remaining schemes under various circumstances. The results of February 3 were taken (Dataset 1) for instance. In comparison with Scheme 4, Scheme 3 enhances the total number of collected data packets by 57.71% when the number of the data center is ten, the speed of code dissemination and the coverage of cabs are ameliorated by 23.92% and 12.93%, respectively. Compared with Scheme 1, the total number of Scheme 3's collected packets is 65.00% higher than that of Scheme 1 when there are ten data centers deployed. The figures for Scheme 3's code dissemination and the coverage of sensing devices exceed that of Scheme 1 by 18.98% and 10.21%, respectively.