Design And Implementation Of An Urban Human Cluster Evolution Mobility Model

Mobile ad-hoc network is inherently cooperative because mobile nodes rely on nearby nodes to deliver messages or maintain network connectivity. Therefor, the underlying behaviors of mobile nodes strongly influence the performance of mobile network. As wireless devices are often carried or controlled by humans, it is important to understanding their behavior patterns to get more realistic network simulation and accurate understanding of the performance of new protocols.Most of the traditional classical synthetic mobility model did not consider the behavior patterns of human beings in the modeling process, and in recent years, due to the limitations of the data set itself, some mobility model which build based on real trace did not consider the features of cluster dynamics which caused by node mobile. In order to solve the issue, this paper study the cluster dynamics during network evolution in human mobile network by analyzing GPS trajectories collected in two different urban environments, and by combining with the random waypoint mobility model, we propose Urban Human Cluster Evolution Mobility Model which reflects the human cluster dynamics features in real urban environment. The main research work includes the following four aspects:I. By summarizing the available trace date, we choose the Shanghai trace and Beijing trace for our study. We analyzed the features of cluster size distribution, reconstruction process of clusters, behavior of free nodes and the lifetime of cluster during network evolution, and we design our mobility model based on these features.II. By study the real traces, this paper proposed the UHCE mobility model, which reflects the human cluster dynamics features in real urban environment.III. The UHCE mobility model was developed by applying C language, which could generate the trace files need by NS-2network simulator.IV. By analyzing the characteristics of UHCE mobility model, comparing the mobility features of cluster dynamics with RWP mobility model and reality, the results show that UHCE mobility model works better in reflecting the cluster dynamics features.