User Mobility Models For Opportunistic Networks

Opportunistic Networking is a novel technique for networking, characterized by its ability of tolerating connection disruption and network partition. In an opportunistic network, there may never be a complete path between the two sides of communication. Messages, in such situations however, can be delivered from its source to its destination, relaying hop by hop with the pattern of store-carry-forward, the progress exploiting opportunities of contact between nodes that are moving continuously. Since an opportunistic network consists exclusively of wireless devices carried by users, user mobility pattern must be considered as a factor of significance in designing the architecture of the network and in evaluating the performance of various protocols. User mobility models contribute more fundamentally and significantly to opportunistic network research than other mobile networks research which only take a nodal mobility model as a basic component in simulators. On the basis of studying a number of nodal mobility models suggested for various mobile networks, and of analyzing characteristics of human movement, in this thesis, a general framework and standard of evaluation of user mobility models for opportunistic networks is proposed. Including 13 features from 5 aspects of human movement behaviors, this framework provides us a comprehensive standard to assess various mobility models, as a guide for us to construct models more coincident to human movement patterns as well. With its guidance, a novel user mobility model, named Purpose-Driven User Mobility Model, is suggested in this thesis. The philosophy of this model is that any human behavior is driven by some purpose, and starting from purposiveness, the model considers together the factors of the correlation between purposes of human being, people's preference to some places and paths, temporal constraints on human movements, the feature of grouping in human movements, and the effect of various random events on human movements. It is a compound mobility model de facto including the two levels of mobility pattern: entity mobility and group mobility. A large number of simulations and analysis show that the proposed mobility model is coincident well to all features of human movement, and we hope it to contribute to other areas of opportunistic network research.