Event-based Production Rules for Data Aggregation in Wireless Sensor Networks

摘要

Traditional sensor networks assume that the sensors are pre-programmed and send their data to a central station where the data is aggregated and analysed. However, energy is a major bottleneck resource in wireless sensor networks, e.g. the lifespan of a node is often determined by its battery lifespan. Since each transmitted bit consumes network power it is necessary to reduce the number of sent messages by pre-processing data. Furthermore scalability issues also demand data processing inside the sensor network. After an introduction and the review of the state of the art with respect to data aggregation, three motivating applications are discussed and then the data aggregation problem is explored by means of event-based production rules. The basic constructs of our aggregation approach are facts, patterns, rules, actions and functions. More energy-rich nodes are used as cluster heads that perform rule-based data processing and aggregation operations within local regions of the network. Rules are usually application specific and provide robust interpretations of sensor readings including the generation of warnings. They are used to encode knowledge and to deal with large sets of rapidly changing data.