Data acquisition and transmission in smart vehicle based community sensing

Abstract

Nowadays, smart vehicles have been increasingly adopted in community sensing applications, because of their ubiquity, mobility and connectivity. Compared with existing community sensing devices (e.g., smartphone), smart vehicles pose several distinctive opportunities and challenges when utilized as community sensing participants. On one hand, vehicles have sufficient energy supply and computation capability for complex data processing, on the other hand, the heterogeneity of wireless communication and highly dynamic mobility hinder the efficiency of data transmission. In this thesis, we investigate new approaches of data acquisition and data transmission in smart vehicle based community sensing applications. First, we study data acquisition techniques in both mobile device based and community sensing based scenario. For mobile device based scenario, we focus on how to acquire the seat level in-vehicle position information of passengers, as it can provide fine-grained context information for many upper layer applications. The proposed approach does not require any extra devices and uses only embedded sensors in smartphones. In-vehicle smartphones automatically collect data during certain events and cooperatively determine the relative left/right, and front/back locations. With complex signal processing, our approach is robust to noisy data and possible sensor errors. We evaluate the performance in real driving scenarios. Results show that the positioning accuracy is promising. For community sensing based scenario, we focus on how to reduce the cost of data acquisition by recruiting appropriate participants. Vehicles have the distinct advantage of predictable mobility, which brings new insight into improving the community sensing quality. We present a new participant recruitment strategy for data acquisition, which guarantees the system performance using the currently recruited participants for a period of time in the future. We propose two algorithms, a greedy approximation and a genetic algorithm, to find the solution for different application scenarios. We also validated the performance of our algorithms with traffic trace dataset. Second, we study data transmission techniques in both decentralized and logically centralized scenario. For decentralized scenario, we utilize real-time traffic information to improve data transmission efficiency. Specifically, we study the data transmission problem in VANETs under different traffic conditions. The objective is to minimize the network communication overhead while satisfying the data transmission time constraint. Our method can adaptively choose to carry or forward data according to current traffic information. Evaluation shows that the proposed solution outperforms some existing ones in terms of effectiveness and efficiency. For logically centralized scenario, we address the challenges in applying software defined network to develop high-performance vehicular networks. We present a new SDN based vehicular network architecture. It organizes the topology of the vehicular networks and utilizes vehicle trajectory prediction to mitigate the overhead of the SDN control and data plane communication. To demonstrate the advantage of the architecture, we propose a multicast protocol over it, which exploits the network topology information provided by SDN to make far more efficient multicast scheduling decision. We conduct extensive experiments using traffic traces. The evaluation verifies the feasibility of the architecture. It also shows that the SDN based multicast protocol outperforms existing decentralized approaches in terms of communication delay, cost and data delivery ratio. Third, we focus on a typical vehicle based community sensing application. We study the problem of how to accurately estimate real-time traffic speed using community sensing. We propose an approach that utilizes macroscopic traffic flow model to estimate the traffic speed. The model only relies on vehicle density, which is less likely to be affected by the temporal dynamicity of traffic. We also develop a hybrid experimental platform using both large scale simulation and small scale testbed for evaluation. The results demonstrate that our approach outperforms some existing ones.