EMM: an event-driven mobility model for generating movements of large numbers of mobile nodes

In many researches related to wireless network, simulation is commonly used for performance evaluation. Among the simulation parameters, the mobility model that is used to generate the movements of mobile nodes (MNs) is essential. Mobility models can be classified into two types: trace or synthetic [Proceedings of the SCS Western Multiconference Web-based Simulation Track, January 1999]. Trace models use traces which show mobility patterns of MNs observed in the real world. Traces provide accurate movements, but they are not easily modeled. On the other hand, synthetic models attempt to realistically represent the movements of MNs without using traces. Most of the current simulations are based on synthetic models. Many applications are designed to serve hundreds or thousands of MNs simultaneously, especially in mobile or pervasive computing. The simulations of such applications require the movements of large numbers of MNs. However, current synthetic models are focus on small numbers of MNs. If they are used to generate movements of large numbers of MNs, the generated movements are not similar to the movements in the real world. It causes the simulation results of the above applications are debatable. Consequently, a trace model, called an event-driven mobility model (EMM), is proposed in this paper. When the real world is observed, it can be seen that the movements of MNs are driven by various events (e.g., 1569-190X/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.simpat.2004.11.005 * Corresponding author. Tel.: +88 6 4 23323000x4211; fax: +886 4 23742375. E-mail address: [email protected] (H.-C. Liao). 336 Y.-C. Chang, H.-C. Liao / Simulation Modelling Practice and Theory 13 (2005) 335–355 meetings, dining, and so on). This allows the movements to be deemed as ‘‘move-stay’’ mobility pattern. A modified colored petri-net is used to model the move-stay pattern. The current design of EMM is based on such pattern of students on a campus. In our experiment, 6780 movements of students in one day were generated systematically. The statistics of the generated movements were used to determine the deployment of wireless access points. The result shows that EMM is useful for generating movements of large numbers of MNs. 2004 Elsevier B.V. All rights reserved.