Modelling Correlated Mobility

TL;DR

This paper introduces a novel model for correlated mobility in mobile networks, using node colouring time to better predict contact patterns and latency, especially in vehicular scenarios.

Contribution

It proposes a new correlated mobility model based on node colouring time and demonstrates its effectiveness over traditional independent contact models.

Findings

Model fits real vehicular mobility traces well.

Correlated contact modeling improves latency predictions.

Independent contact models are less accurate.

Abstract

When nodes in a mobile network cluster together or move according to common external factors (e.g., cars that follow the road network), the resulting contact patterns become correlated. In this work we address the question of modelling such correlated mobility movements for the analysis of intermittently connected networks. We propose to use the concept of node colouring time to characterise dynamic node contact patterns. We analyse how this model compares to existing work, and demonstrate how to extract the relevant data from actual trace files. Moreover, we show how this information can be used to derive the latency distribution of DTN routing protocols. Our model achieves a very good fit to simulated results based on real vehicular mobility traces, whereas models which assumes independent contacts do not.