An event detection framework for wireless sensor networks
2017-03-01T00:05:00Z (GMT) by
Wireless Sensor Networks (WSNs) introduce a new paradigm for sensing and disseminating information from various sources in the physical environment around us. WSNs facilitate the detection of various real-world phenomena (e.g. environmental anomaly, natural disasters, structural faults, man-made hazards and so forth) and thereby aim to reduce any economic and human loss. Typically, a WSN consists of a large number of sensor nodes deployed over a geographic area and each sensor is capable of sensing one or more attributes of the surrounding environment. Unlike traditional communication networks, the structure of a WSN is tightly coupled with the target application. Therefore, the design of a WSN based event detection system involves careful consideration of application-specific performance requirements, diversity in real-world sensing fields, impacts and contexts of the types of target events. The research presented in this thesis focuses on the reliability and accuracy of the detection of physical events using WSNs. Studies in the relevant literature suggest the use of node redundancies in the form of k-coverage to ensure robust detection. In this thesis, first, an optimal QoS support framework for k-coverage in an event-centric WSN using static nodes is presented. Then the concept of coverage hole recovery using variable range sensing is introduced that deals with the loss of coverage arising from node faults and ageing in the post deployment scenario. However, the redundant coverage can be cost prohibiting as the number of nodes becomes very high for high degree of coverage requirement. To reduce the deployment cost, a dynamic k-coverage scheme is proposed that ensures 1-coverage during deployment and provides k-coverage on-demand only after an event is sensed by at least one node. Two different solutions - one using the sensing range adjustment technique and another using node mobility are presented and compared for performance and cost analysis. The proposed detection scheme in this thesis also considers the detection of multiple simultaneous events in a WSN where events may have different priorities depending on their locations and costs of missed-detections. The differentiated treatment for event on a priority basis is evident in many real-world applications. Finally, incorporating the context information in conjunction with the sensed data extends the event detection framework and facilitates the seamless integration of WSNs to the Internet of Things for event detection purpose.