Title:Opportunistic Selection of Vehicular Data Brokers as Relay Nodes to the Cloud

Abstract:The Internet of Things (IoT) revolution and the development of smart communities have resulted in increased demand for bandwidth due to the rise in network traffic. Instead of investing in expensive communications infrastructure, some researchers have proposed leveraging Vehicular Ad-Hoc Networks (VANETs) as the data communications infrastructure. However VANETs are not cheap since they require the deployment of expensive Road Side Units (RSU)s across smart communities. In this research, we propose an infrastructure-less system that opportunistically utilizes vehicles to serve as Local Community Brokers (LCBs) that effectively substitute RSUs for managing communications between smart devices and the cloud in support of smart community applications. We propose an opportunistic algorithm that strives to select vehicles in order to maximize the LCBs' service time. The proposed opportunistic algorithm utilizes an ensemble of online selection algorithms by running all of them together in passive mode and selecting the one that has performed the best in recent history. We evaluate our proposed algorithm using a dataset comprising real taxi traces from the city of Shanghai in China and compare our algorithm against a baseline of 9 Threshold Based Online (TBO) algorithms. A number of experiments are conducted and our results indicate that the proposed algorithm achieves up to 87% more service time with up to 10% fewer vehicle selections compared to the best-performing existing TBO online algorithm.