Mobility and Traffic Correlations in Device-to-Device (D2D) Communication Networks

The development of wireless networking technologies has brought tremendous changes to people's everyday life, such as cellular and WiFi Internet access. It empowers users to communicate through different wireless access technologies, and also enables new applications in new paradigm, such as vehicular ad hoc networks (VANETs) and mobile cloud computing (MCC). Wireless communication can be achieved through not only infrastructure wireless networks (i.e., cellular and WiFi networks) but also opportunistic device-to-device (D2D) communications. One of the major features of D2D communication networks is user mobility, which affects network connectivity and the design of network protocols. Another important feature of D2D communication networks is complex traffic flow due to dynamic network topology and the double communication opportunities (i.e., infrastructure or infrastructureless wireless networks).;In this dissertation, we aim at understanding the mobility and traffic correlation in D2D communication networks, especially the emerging wireless applications such as VANETs and MCC. We first characterize node mobility and mobility correlation among users so that we can identify the autonomous ad hoc networks. Based on observations from real mobility traces, we define a metric, namely Dual-Locality Ratio, to quantify mobility correlation and evaluate group structures. As correlated mobility leads to presence of groups in which nodes have unequal abilities to relay data to other parts of the network, we further study how the information propagates in VANETs, which have highly dynamic and correlated vehicle mobility due to road layout and speed limit. We derive the farthest distance that message dissemination reaches at time t and the first time that message reaches distance d from the original source location under different dissemination strategies. Our analytical bounds provide not only spatial and temporal limits of message dissemination but also guidelines for design of message dissemination algorithms. Recently, D2D communication network is also used to accommodate mobile cloud computing. Besides access remote cloud through cellular or WiFi networks, users can employ nearby mobile devices for mobile cloudlet computing. In order to find out whether the traffic of computation offloading goes to the remote cloud or mobile cloudlet, we address the issue of whether/when mobile cloudlet can provide mobile application services by investigating its properties and computing performance. Finally, we investigate the content delivery in the D2D communication networks such as to accommodate the explosive mobile traffic. We find out how likely D2D communications can deliver contents to mobile users through discovering content distribution in network. The work in this dissertation advances our understanding of mobility and traffic correlation and offers guidance into the design of D2D communication networks.