Scalable mobile data streaming with trajectory preserving partitioning

Abstract

With the rise of sensor technology, ubiquitous mobile devices generate high-rate spatio-temporal data streams and send to backend servers in real time. While the mobile data streams are critical sources of mobility analytics supporting advanced mobile services, they also raise new challenges of scalable data processing for trajectory-featured data streams. In this paper, we develop a scalable framework for mobile data streaming and present its architectural strategy. By introducing the mobility localization theme for the core partition component, a geo-spatial partition method is proposed for high performance data stream dispatching. Experiments on real world data demonstrate the effectiveness of the proposed method.