Variety event detection in Wireless Sensor Networks through single hop cluster topology

Abstract

Event detection techniques in Wireless Sensor Networks (WSNs) are important for environment monitoring and object tracking. Here, we focus on Single hop WSNs since they have received very little attention from researchers, with extremely few contributions being available, although the usefulness of these single hop WSNs are widely accepted. Furthermore, in this paper, we achieve the necessary requirement of establishing the effectiveness of proposed WSN protocol purely through theoretical analysis, thus avoiding the need for incomplete and multiple error-prone simulation results, which can form the basis of some important results in later contributions. We have proposed here a new protocol named SCHTED on single hop cluster tree topology for multiple and variety event detection in WSNs. In this model, a WSN is divided into several clusters. The clusters contain high but varied density of sensors. Each cluster contains a cluster head which is a child node of some other cluster head. In this way the entire network is arranged in a tree topology with the root node as the sink. Few sensors in each cluster, randomly selected by the sink, sense and measure a predefined set of parameters from the environment. Thereafter, the sensors compare their current measurement with that of the previous sensing step. If any sensor finds significant difference in the two measurements, it multicasts the information of occurrence of an event to its neighboring cluster members. If the number of sensors sensing a significant difference in measurements in consecutive sensing step exceeds a certain predefined threshold, as detected by the cluster head, then the cluster head send notification to its neighboring cluster head that an event has occurred. This information travels through the tree topology to ultimately reach the sink. Different cases of event detection such as sensitive events, severe events, high frequency events, real time processing, point event as well as region events are considered here. Analysis reveals that this protocol scheme is suitable for both continuous and event based monitoring in noisy environment and can also be made adaptive to the changing requirements.