An Efficient Group Partition Prediction Scheme for MANETs

Group mobility is prevalent in many MANET applications, such as disaster recovery, military actions, etc., and group partitions are unavoidable under such dynamic network. Group partitioning may result in frequent link disconnections, which hence affects the inter-group communication. In this paper, we introduce a framework for predicting potential group partitions in MANETs such that remedial actions can be taken in advance. The weighted linear prediction model is proposed in our work which is based on the linear displacement of a group instead of its instant velocity. The concept of safe distance is discussed and it is used in predicting the partition time of adjacent groups. The experiment shows that the prediction model displays its efficiency in prediction of group partitions.