Toward stochastic anatomy of inter-meeting time distribution under general mobility models
Han Cai, Do Young Eun · 2008
Recent discovery of the mixture (power-law and exponential) behavior of inter-meeting time distribution of mobile nodes presents new challenge to the problem of mobility modeling and its effect on the network performance. Existing studies on this problem via the average inter-meeting time become insufficient when the inter-meeting time distribution starts to deviate from exponential one. This insufficiency necessarily leads to the increasing difficulty in the performance analysis of forwarding algorithms in mobile ad-hoc networks (MANET). In this paper, we analyze the effect of mobility patterns on the inter-meeting time distribution. We first identify the critical timescale in the inter-meeting distribution, at which the transition from power-law to exponential takes place, in terms of the domain size and the statistics of the mobility pattern. We then prove that stronger correlations in mobility patterns lead to heavier (non-exponential) 'head' of the inter-meeting time distribution. We also prove that there exists an invariance property for several contact-based metrics such as inter-meeting, contact, inter-any-contact time under both distance-based (Boolean) and physical interference (SINR) based models, in that the averages of those contact-based metrics do not depend on the degree of correlations in the mobility patterns. Our results collectively suggest a convex ordering relationship among inter-meeting times of various mobility models indexed by their degrees of correlation, which is in good agreement with the ordering of network performance under a set of mobility patterns whose inter-meeting time distributions have power-law 'head' followed by exponential 'tail'.