Group Mobility: Detection, Tracking and Characterization

TL;DR

This paper investigates human mobility from a social perspective by analyzing group behaviors, periodicity, and dynamics using proximity data, with implications for routing and information dissemination in mobile networks.

Contribution

It introduces a social group-based analysis of mobility patterns, revealing periodicity and dynamics that can improve predictive algorithms and protocols.

Findings

Group meetings are periodic daily and weekly.

Longer group meetings have less member change.

Group dynamics vary over different times of the day.

Abstract

In the era of mobile computing, understanding human mobility patterns is crucial in order to better design protocols and applications. Many studies focus on different aspects of human mobility such as people's points of interests, routes, traffic, individual mobility patterns, among others. In this work, we propose to look at human mobility through a social perspective, i.e., analyze the impact of social groups in mobility patterns. We use the MIT Reality Mining proximity trace to detect, track and investigate group's evolution throughout time. Our results show that group meetings happen in a periodical fashion and present daily and weekly periodicity. We analyze how groups' dynamics change over day hours and find that group meetings lasting longer are those with less changes in members composition and with members having stronger social bonds with each other. Our findings can be used to propose meeting prediction algorithms, opportunistic routing and information diffusion protocols, taking advantage of those revealed properties.