Event-driven query processing in SNQL for sensor database

Abstract

With rapid development of sensor devices and wireless network technologies the existing applications using sensor data have become ever more sophisticated and also a variety of new information services are emerging for the future computing services. Many researches and development called for data collection and management facility over the sensor networks in ways similar to the traditional database management. Different from the traditional database systems the sensor data management requires querying the sensor nodes instead of stored data. In the sensor database management, queries are disseminated from the base station to the sensor nodes where they are executed to collect sensory data and transmit them back to the base station. Among many particular features is the expression and processing of event-driven queries that are triggered and executed when some events are detected at sensor nodes. Therefore, applications don`t have to monitor the dynamic occurrence of the events in the sensor network. Instead, event queries are disseminated to the sensor nodes and let them watch the events. When detected the event node(s) propagate the queries toward the designated sensor nodes where the queries are executed. Some of sensor database systems support such event-driven queries, but they have limited language expressions to efficiently implement complex application requests. In this paper, we propose an event-driven query language facility and processing algorithms as an improvement of our previous sensor database query language, Sensor Network Query Language (SNQL). The proposed event-driven queries provide expressions, such as temporal condition of monitoring the event, selective query dissemination over the network with a simple spatial operations, and in-network propagation of inner query when events are triggered. For the management of selective dissemination and in-network propagations, we have implemented a Quadtree-based Distributed MBR Management (QDMM) ...