MEME: Real-time mobility estimation for mobile environments

Abstract

Knowledge of user movement in mobile environments paves the way for intelligent resource allocation and event scheduling for a variety of applications. Existing schemes for estimating user mobility are limited in their scope as they rely on repetitive patterns of user movement. Such patterns may not exist, or are difficult to recognize in soft-real time, in open environments such as parks, malls, or streets. We propose a novel scheme for Real-time Mobility Estimation for Mobile Environments (MEME). MEME employs the concept of temporal distances and uses logistic regression to make real time estimations about user movement. MEME relies only on opportunistic message exchange and is fully distributed, scalable, and requires neither a central infrastructure nor Global Positioning System. MEME has been tested on real world and synthetic mobility traces - makes predictions about direction and count of users with up to 90% accuracy, enhances successful video downloads in shared environments.