Abstract

Abstract. Continuously tracking mobility of humans, vehicles or merchandise not only provides streaming, real-time information about their current whereabouts, but can also progressively assemble historical traces, i.e., their evolving trajectories. In this paper, we outline a framework for online detection of groups of moving objects with approximately similar routes over the recent past. Further, we propose an encoding scheme for synthesizing an indicative trajectory that collectively represents movement features pertaining to objects in the same group. Preliminary experimentation with this multiplexing scheme shows encouraging results in terms of both maintenance cost and compression accuracy. Motivation As smartphones and GPS-enabled devices proliferate and location-based services penetrate into the market, managing the bulk of rapidly accumulating traces of objects' movement becomes all the more crucial for monitoring applications. Apart from effective storage and timely response to user requests, data exploration and trend discovery against collections of evolving trajectories seems very challenging. From detection of flocks We have begun developing a stream-based framework for multiplexing trajectories of objects that approximately travel together over a recent time interval. Our perception is that a symbolic encoding for sequences of trajectory segments can offer a rough, yet succinct abstraction of their concurrent evolution. Taking advantage of inherent properties, such as heading, speed and current position, we can continuously report groups of objects with similar motion traces. Then, we may regularly construct an indicative path per detected group, which actually epitomizes spatiotemporal features shared by its participating objects. Overall, such a scheme could be beneficial for: -Data compression: collectively represent traces of multiple objects with a single "delegate" that suitably approximates their common recent movement. -Data discovery: find trends or motion patterns from real-time location feeds. -Data visualization: estimate significance of each multiplexed group of trajectories and illustrate its mutability across time (e.g., on maps).