Using statistical data for reliable mobile communications

We examine the extent to which statistical mobility information can increase the reliability of the service experienced by users in mobile networks. Interrupted or dropped calls are an aspect of reliability that stems from the mobility of users. An existing user can move to a cell where there are no resources available to support their call. A natural solution is the reservation of resources in multiple cells that the user is likely to move to. This scheme is called selective reservations and it relies on predicting the next move of the user. Recently, there has been some work on estimating the movement probabilities (also known as the mobility profile) of the user. In this paper, we quantify the usefulness of the mobility profile to improve the reliability of the service perceived by the mobile users. We identify two parameters which characterize the profile: Accuracy and Focus. Accuracy expresses the probability that the host will move as we expect it to. Focus describes how well we can identify patterns in the movement of the users. In our simulations, we examine the effect of the quality of the predictions on the performance of the system. We show that Accuracy and Focus have great impact on the performance of selective reservations. We also show how flexibility in hand-offs can help in decreasing the dropping probability, and how this can be facilitated by letting the users make a second try at moving in case it fails the first time.