Decentralised cooperative agent-based clustering in intelligent traffic clouds

Abstract

Contemporary traffic management systems will become more intelligent with advent of future Internet technologies. The systems are expected to become more simple, effective and comfortable for users, but this transformation will require the development of both new system architectures as well as enhanced processing and mining algorithms for large volumes of cloud data. In this study, we consider a conceptual architecture of a cloud-based traffic management system that applied to a multi-modal journey planning scenario. For this purpose, it is necessary to process large amounts of travel-time information. Information is collected by cloud service providers and processed for future route planning. In this paper, we focus on the data clustering step in the data mining process. The data collection and processing require an appropriate clustering algorithm to aggregate similar data. In particular, we support a process where a particular service provider can request additional information from others to be used in the clustering function, requiring a decentralised clustering algorithm. We present a cloud-based architecture for this scenario, develop a decentralised cooperative kernel-density based clustering algorithm, and evaluate the efficiency of the proposed approach using real-world traffic data from Hanover, Germany. © 2013 Springer-Verlag Berlin Heidelberg.