I An Effective Approach for Mobility Prediction in Wireless Network based on Temporal Weighted Mobility Rule

In the wireless network, whenever a mobile node moves from one cell to another called handover or handoff, the call needs to be handed off to the new base station, and then network resources must be reallocated. Many mobility prediction schemes are proposed to perform resource reservations in advance so as to reduce the handover latency. Such approaches make use of knowledge patterns of location being mined from the mobility history of users to describe and predict the movement of mobile users. In addition to the location characteristic, the timeof-day also plays a crucial role in modeling the movement and it has attracted several research interests recently. In this paper, we investigate simultaneously spatial and temporal attributes of data and apply a spatiotemporal data mining technique to discover frequent mobility patterns for predicting the next location of a mobile node. Our approach is to mine frequent mobility patterns with time and then to make use of them to construct temporal weighted mobility rules. This paper extends a mobility prediction algorithm by finding the best matched rules which are temporally closest to the query time. Our experimental results show that using the temporal attribute is necessary for improving the prediction accuracy.