Mobility Models for Delay-Tolerant Mobile Networks

In this paper, we investigate sophisticated mobility models that can be useful for researches in Delay-Tolerant Mobile Networks. Our research is based on the observation of existing movement traces with several simplifications for network tractability. First, we present a continuous Markovian mobility model, where mobile nodes stay at one location or move from locations to locations according to exponential time distributions. This assumption has been widely adopted in many analyses and simulations. Then we discuss a discrete Markovian mobility model,where a node makes movement decisions according to given probabilities. We prove the discrete model is equivalent to its continuous counterpart. They both have high complexity due to the potentially large number of states. To reduce the complexity, we present a Community-based mobility model, where each node has its own home location and moves in or out that location according to given probabilities. The Community-based model is a special case of the discrete Markovian model with a much smaller set of states and matches well with the real user traces. Based on Markov-chain theory, we can extract the stationary state probabilities from these models, which are useful for many analytical and simulation studies. Extensive simulations have been carried out to validate our discussions.