Event driven querying of semantic sensor web services
Metadata[+] Show full item record
In today's world, there is a tremendous increase in the usage of sensor technology in several fields including agriculture, medicine, and weather. Sensors either in-site or remotely placed are usually deployed in the form of sensor clusters containing dozens of sensor nodes. The observed data from the various sensors need to be thoroughly analyzed for understanding the content they carry. Human analysis of the observed data is not always feasible or convincible, and hence, although there has been a rapid growth of sensor technology, the practical applications are far from reality. Therefore, there has been an imminent need for an intelligent approach to understand the data from the sensor clusters. At the same time, controlling the amount of the observation data is also necessary. To address this integral issue, we present the Event Driven Querying of Semantic Sensor Web Services. The Event Condition Action (ECA) based model is intended for providing a platform for querying the cluster of sensors in an efficient and timely manner. Processing observation data is surpassed by semantically annotating data and using rule based reasoning as an inference tool. ECA enables a shift of the main focus from a large cluster section to a precise smaller section of the cluster, and thus eliminates the necessity to obtain data from entire sensor cluster every time to make an inference. This model can be used to measure and track numerous events like earthquakes, floods, stock market crashes, Christmas shopping trends that are set off by a pre-condition that in turn triggers a set of events. In order to validate the efficiency and preciseness of the proposed model, we introduced an example of detection and prorogation of fire in a closed two dimensional building. The interior rooms of the building are modeled as sensor nodes for maintaining a state. We found the occurrences of certain events, like rise in temperature and production of smoke using proactive approach, were the forerunners of fire in the room. These events are captured by the reasoning engine from the data obtained from the sensors, resulting in a change of state in the sensor nodes. The change then triggers new events, bringing about a cascading waterfall like effect. Finally, we measured the accuracy of the model by considering a sensor networks consisting of 11, 25, 50 and 100 nodes with up to two sources of fire and while tracking its propagation. We also measured the F-measure for each of the above sensor networks. Thus, the ECA model based coupled with proactive querying helps in not only curtailing the amount of observation data, but also helps in accurately determining the source of an event and tracking its spread effectively.
Table of Contents
Introduction -- Related work -- Framework for event driven querying of semantic sensor web services -- Architecture of event driven querying of sensor web services -- Evaluation -- Conclusion and future work