Theoretic studies on delay-tolerant mobile networks: a queuing theory approach

In this dissertation, we introduce theoretic foundations of delay-tolerant mobile networks, including mobility and analytic models. We first introduce three sophisticated mobility models, which are designed based on human movement patterns, namely a continuous Markovian mobility model, a discrete Markovian mobility model, and a community-based mobility model. We then introduce two analytic queuing models for delay-tolerant mobile sensor networks and delay-tolerant mobile ad hoc networks. The first model is developed based on the analysis of the message arrival and service processes at each individual mobile sensor. Consequently, the performance of the whole network, such as the average number of data messages in the system, the average queuing delay of each message, and the network-wide message delivery ratio, can be obtained. We then extend our analysis to peer-to-peer delay-tolerant mobile ad hoc networks by using a multi-dimensional Markov-chain model. Although a potentially large number of states and state equations must be dealt with in the Markov chain, a closed-form solution is found in our studies. We also develop an approximation solution in order to support the analysis of very large delay-tolerant mobile ad hoc networks. In both considered networks, we observe the strong impact of queue length and nodal density on network performance, due to the unique characteristics of such networks with extremely low connectivity. We expect the proposed queuing models will give insights into the queuing behavior of delay-tolerant mobile networks and provide useful guidelines for future protocol design and optimization. Finally, we propose a cluster-based data transmission protocol for delay-tolerant mobile networks. The protocol is proven to be effective in cluster-oriented mobility patterns such as the aforementioned community-based mobility model.