Mobility Modeling, Analysis, and Applications in Challenged Networks

Summary: We describe our main research topics on social activity based mobility modeling, encounter based mobility analysis and mobility-assisted networking applications.

1 Social activity driven mobility modeling

Among most existing mobility models, a largely missed issue is human's social roles in making movement decisions, which if modeled properly, would make a more realistic network scenario. To tackle the problem, we developed an Agenda Driven Mobility Model. The model is a framework that takes a person's social activities in his or her agenda (when, where and what) and generates mobility in geographic movements. We use a constructive approach to define functional components of the model for building specific real world scenarios and generating motion steps, which translating agendas into a mobile world. Various sources of real data can be used to populate these components, including GIS, consensus on human travel and wireless trace data. Thus far, we are able to create urban mobile environment through using National Household Travel Survey (NHTS) data in our model. Through extensive study our model allows us to observe the uneven distributions of node geographic concentrations, as a demonstration of nodes' social roles and activities. Our results further show high dynamics about node concentration, connectivity and partition in both spatial and time domains. We found that the incorporation of social roles and agenda activities into mobility modeling has significant impact on routing protocol performance. The model has added significant capability and flexibility to the current mobility modeling research area through capturing key properties of a real mobile node’s social roles and collectively making them to represent a mobile network environment.

Qunwei Zheng*, Xiaoyan Hong, Jun Liu*, Wan Huang, David Cordes, "Agenda Driven Mobility Modeling",  International Journal of Ad Hoc and Ubiquitous Computing, accepted, 2008.

Qunwei Zheng*, Xiaoyan Hong, Jun Liu*, “An agenda based mobility model,” 39th Annual Simulation Symposium, Huntsville, AL., April, 2006.

Qunwei Zheng*, Xiaoyan Hong, Sibu Ray, ``Recent Advances in Mobility Modeling for Mobile Ad Hoc Network Research’’, Proc. 42nd Annual ACM Southeast Conference, Huntsville, Alabama, USA, April 2-3, 2004, pp 70-75.

2 Encounter analysis and delay management for DTN

In delay tolerant networks (DTN), nodes with a message to transmit rely on encountering other nodes during movements and exchanging the message with each other. A series of encounters of different nodes will spread the message among many nodes and eventually deliver it to the designated destination. The quality of the message delivery heavily depends on the probability of two nodes meeting each other (encounter probability), which indicate delivery success rate; and the time it takes for two nodes to meet (encounter delay). Some critical factors influence these properties greatly, including the way a node picks its next location (mobility patterns) and the features of the dwell time. We have made unique contributions by analyzing these properties as functions of time using discrete and continuous time Markov Chain models. Our analysis consider many constraints including random, or semi-random (agenda-based) location selection, constant and time-variant dwelling time. We also analyzed a constraint on the number of locations a message carrier will visit. The results help generating the desired delivery curves and selecting proper time-to-live thresholds for a desired delivery probability. The methodology we use can extend to other possibly more complicated scenarios and mobility patterns.

Further, we have exploited location visit properties in an efficient routing scheme for DTN called SMART. The SMART uses a combination of large fan out and focused message forwarding in order to limit message overhead and to improve the delivery ratio. The encounter probability and latency analysis helps a node to learn and to select the nodes that frequently meet the destination to achieve the design goals. Our analysis and simulation results show that SMART has a higher delivery ratio and a smaller delivery latency than the controlled opportunistically-forwarding schemes and has a significantly smaller routing overhead than the pure flooding schemes.

Qunwei Zheng*, Xiaoyan Hong, Pu Wang, Lei Tang*, Jun Liu*, “Delay Management in Delay Tolerant Network,” Special Issue of International Journal of Network Management on Mathematical Methods in Network Management, accepted, 2008.

Qunwei Zheng* (May 2007), dissertation: “Mobility in Mobile Ad Hoc Networks: Models, Analysis, and Applications.” University of Alabama.

Lei Tang*, Qunwei Zheng*, Jun Liu*, Xiaoyan Hong, “Selective Message Forwarding in Delay Tolerant Networks,” Mobile Networks and Applications, Special Issue on Broadband Communications, Networks, and Systems, accepted, 2008.

Lei Tang*, Xiaoyan Hong, Qunwei Zheng*, Jun Liu*, “SMART: A Selective Controlled-Flooding Routing for Delay Tolerant Networks”, International Conference on Broadband Communications, Networks, and Systems (Broadnets 2007), Raleigh, NC, Sept 2007.

3 Mobility applications: security for delay tolerant applications

We further explored mobility to disseminate message securely in mobile ad hoc networks. The multipath approach, i.e, sending many secret shares of the original message through multiple different paths, has been used to deal with the challenge of establishing strict security mechanisms in the self-organizing mobile networks. One major limitations of the spatial multi-path scheme come from the scenarios when a network is sparse or deployed in a geographically constrained area, where not enough node-disjoint multiple paths can be found. To tackle the issue, we developed a novel protocol that explores multiple routing paths in time domain for applications that can tolerate delay. In our scheme, the source sends shares at different times. Due to node mobility, the shares will be routed through different intermediate nodes. Using analysis and worse-case simulation, we proofed that it is high unlikely that a particular intermediate node (an eavesdropper) is able to be on many of these routes and to collect enough shares to reconstruct the original message. Of course, the sender can control its sending intervals to achieve a high security.

Qunwei Zheng*, Xiaoyan Hong, Jun Liu*, lei Tang*, “Secure Data Transmission Scheme for Mobile Ad Hoc Networks”, in the Proceedings of IEEE Global Telecommunications Conference (Globecom), Washington, D.C., USA, Nov 26-30, 2007.