Highly mobile users and paging: optimal polling strategies

We consider the problem of minimizing average paging cost subject to delay constraints in a wireless system. Previous work assumed the unit to be found did not move during the paging process whereas here the unit may change location during polling events. We show that the conditionally most probable locations, given that the unit has not yet been found, should be searched first. We find the optimal sequential paging strategies for given maximum delay constraints and compute both paging and delay costs as a function of the time between polling events. The results show that sequential paging strategies are beneficial in all but the extremely high-mobility cases where polling failures provide little information about the unit location. It is observed that optimal sequential paging strategies substantially lower the paging cost compared to the classical blanket polling at the expense of a small degradation in the average delay performance.