Parameterized Spatio-Textual Publish/Subscribe in Road Sensor Networks

Abstract

Huge amounts of data that are geo-tagged and associated with text information are being generated at an unprecedented scale in road sensor networks. Publish/subscribe system is one kind of important applications for analyzing and processing these huge mounts of data in road sensor networks, which is required to support millions of subscriptions and filter a message in milliseconds. Since the messages arrive continuously at a high speed, rapid processing of the messages is definitely a challenge. This paper mainly addresses the issue of parameterized spatio-textual publish/subscribe problem in road sensor networks. First, with considering both the network distance and textual similarity of the subscriptions and messages, the road network structure, together with the subscriptions and the messages will be partitioned and organized efficiently, and a combined index structure, called basic indexing architecture, is proposed. Second, several effective pruning techniques which consider both location information and textual information are presented to cut down the processing overhead. Moreover, by employing these pruning techniques into the basic indexing architecture, an more efficient index, called enhanced indexing architecture, is presented. Third, an efficient processing algorithm is designed to improve the scalability. Finally, extensive simulations are conducted to show the efficiency and scalability of the proposed methods in road sensor networks.