Model And Applications In Opportunistic Mobile Networks

With the popularization of personal hand-held mobile devices (e.g., smartphones, MP3 players, and PDAs), Opportunistic Mobile Network (OppNet) is arising recently. Without the constraint of real-time connectivity, OppNet is a more flexible ad hoc net-work, and has significant impact on future pervasive computing. It can be widely applied in hand-held device networking, wildlife monitoring, vehicular networks, and rural com-munication, etc. Being able to improve people's productivity and lives, OppNet has gained high attention from global researchers.Thanks to the hand-held device networking paradigm, the characteristics of OppNet mostly depend on people's mobility. Therefore, OppNet can be viewed as a mobile social network, and the use of Social Network Analysis (SNA) will definitely facilitate the re-search on OppNet. Based on the latest results, this dissertation studies the modeling and several application schemes in OppNet from a social network perspective. The novelty and contributions are summarized as follows:1. A brief literature review on the development of OppNet and related work is provided.2. Exploring several realistic datasets to reveal both geographic and social regularities of human mobility, and further proposing the concept ofgeo-community into OppNet analysis. A semi-Markov process is then employed to model node mobility based on the geo-community structure of the network.3. The issue of active data broadcasting is studied. The objective is to broadcast data from a supernode to other nodes in the network. The geo-centrality indicating the "dynamic node density" of each geo-community can be derived from the semi-Markov model. Then, several route algorithms are provided considering the geo-centrality information. 4. A novel two-hop delegation query (DelQue) scheme is presented, which considers query and response integratedly. In contrast to other existing multi-hop search ap-proaches, DelQue can lower energy cost dramatically. Furthermore, a spatio-temporal prediction method of node mobility based on semi-Markov model is also proposed to compute neighbors'query utility. Such a lightweight forcasting technique only requires nodes to maintain two parameter matrices, making it suitable for a resource-scarce mobile setting.5. An incentive-based content subscribe (ConSub) scheme is proposed in selfish Opp-Nets. Interests subscribed and content published by network nodes both are classified into channels. Data dissemination in the network is driven by interest matching. Dur-ing data exchange between two nodes, TFT is employed to prevent "fully-detect" or "free-ride" phenomenon. A trading scheme is implemented in ConSub that motivates each mobile node to act as a merchant and collect in its buffer an inventory of data that maximizes its future trading capacity.6. Trace-driven simulations show that the above modeling and application schemes consistently outperform other existing approaches.The conclusions and future work are depicted in the end of the dissertation.