Understanding Urban Dynamics with Big Data on Mobile Traffic

Abstract

Understanding mobile data consumption patterns is crucial for learning about urban ecosystem and human activity. This task is challenging in the sense that complexity of mobile data usage in vast metropolitan environments, the disruption of unusual events, and the absence of prior understanding of urban traffic patterns are the three problems. We suggest a fresh method for creating a strong system that consists of three subsystems: time series decomposition of mobile traffic data, pattern extraction from various elements of the original traffic, and anomalous event detection from noises. Three significant findings come from our examination into the mobile traffic data of 6,400 cell towers in Shanghai. First, we find five daily patterns among the 6,400 cellular towers that correlate to various human daily activity patterns.