Skip to Main Content
We propose and analyze a class of autonomous host-centric mobility prediction algorithms via simulation for predicting future movements of mobile hosts based on recent and past movement histories observed. These mobility prediction algorithms are 'autonomous' and 'host-centric' since they require individual mobile hosts, rather than fixed base stations, to collect and maintain mobility prediction data, thus making them applicable to both cellular and ad hoc wireless networks. Each mobile host under these autonomous host-centric mobility prediction algorithms is capable of predicting the probability that it will leave the current location to the next location within a time period, thus facilitating resource reservation and route maintenance decisions, as well as for estimating the mobile host's residence time. We describe a simulation environment for evaluating the prediction accuracy for this probability metric by these algorithms, and present the evaluation results with physical interpretations and their applicability given.