Design, Modeling, and Analysis of User Mobility and its Impact on Multi-hop Wireless Networks

Show full item record

Title: Design, Modeling, and Analysis of User Mobility and its Impact on Multi-hop Wireless Networks
Author: Zhao, Ming
Advisors: Arne A. Nilsson, Committee Member
Wenye Wang, Committee Chair
Khaled Harfoush, Committee Member
Michael Devetsikiotis, Committee Member
Abstract: Due to the readily deploy-able and self-organizing nature of mobile ad hoc networks (MANETs), various demanding applications are expected to be widely performed in military, civilian,law enforcement, and disaster recovery environments etc. Since MANETs do not rely on fixed wireless infrastructures, the constrained power consumption of wireless devices limits the communication zone between mobile nodes, so that the communication between nodes has to via a multi-hop fashion. The fundamental issue in MAENTs is that performance degrades dramatically as the increase of path failures. A path failure occurs when the inter-media nodes along the path become unavailable due to the node mobility. Hence, node mobility is a dominating factor to MANET performance and in turn, the most fundamental problem in mobility related MANET studies. The study of node mobility in MAENTs has drawn great attentions in the past decade, nevertheless the in-depth interpretation of mobility patterns and their effects upon system results still remains elusive. Therefore, understanding the impacts and implications of node mobility is essential to the system design, topology control, and routing optimization in MANETs. In this doctoral study, we first propose a novel mobility model named Semi-Markov Smooth (SMS) model, after finding that existing random mobility models can lead to biased or even misleading results for MANETs performance evaluation and analysis. We showed that the proposed SMS model not only captures the transient user mobility according to the physical law of a smooth movement, but also achieves the necessary stationary properties including stable average speed and uniform node distribution. We find that MANET link performance is influenced by the joint effects of radio channel environments and node mobility. Hence by applying the results from our first work, we continue to investigate the link properties upon these two joint effects in the second work. We showed that radio channel characteristics predominate the link performance for slower mobile nodes, while node mobility dominates the link performance for faster mobile nodes. In addition, the link lifetime distribution can be effectively approximated by the exponential distribution with parameter V/R, characterized by the ratio of average node speed to the transmission range. As wireless devices are generally attached to humans, in the third work, we studied human mobility impact on the properties of contact-based metrics, especially the inter-meeting time (MANETs). By investigating the empirical human mobility traces, we demonstrated that both pause time and trip displacement of human mobility exhibit a cutoff-power distribution. Then, we showed that the human diffusive rate r characterizes the joint temporal-spatial effect of human mobility regarding pause time and trip displacement on inter-meeting time. Especially, when the power law head characterizes the distribution of human mobility, we showed that the human diffusive capability (rate) r is, r = 2α/β, where α and β are the power law coefficients for the pause time and trip displacement and 0 < α < 1 and 0 < β < 2. Upon simulation demonstration, we find that the human diffusive rate r directly captures the cutoff power law property of inter-meeting time. Especially, the higher the diffusive rate r is, the shorter the power law head is, while the longer the exponential tail becomes in the distribution of inter-meeting time. In the fourth work, we studied the inherent properties of group mobility in MANETs. Specifically, we analyzed the correlation between group members by taking human social correlation, similarity of the nodes’ movements and proximity of nodes’ geographical location into consideration. We observed that human groups in MANETs exhibit the group evolution process regarding the variation of group size (the total number of group members) with time. Hence, based on the group correlation metric, we further studied several metrics for characterizing the stability of group structures and investigated the condition of node switch between neighboring groups. Finally, by applying the studied metrics, we propose a novel birth-to-death group mobility to describe the group evolution behaviors in MAENTs. Electrical and Computer Engineering
Date: 2009-08-05
Degree: PhD
Discipline: Computer Engineering

Files in this item

Files Size Format View
etd.pdf 2.892Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record