A novel energy efficient wireless sensor network framework for object tracking

Object tracking is a typical application of Wireless Sensor Networks (WSNs), which refers to the process of locating a moving object (or multiple objects) over time using a sensor network. Object tracking in WSNs can be a time consuming and resource hungry process due to factors, such as the amount of data generated or limited resources available to the sensor network. The traditional centralised approaches where a number of sensors transmit all information to a base station or a sink node, increase computation burden. More recently static or dynamic clustering approaches have been explored. Both clustering approaches suffer from certain problems, such as, large clusters, redundant data collection and excessive energy consumption. In addition, most existing object tracking algorithms mainly focus on tracking an object instead of predicting the destination of an object. To address the limitations of existing approaches, this thesis presents a novel framework for efficient object tracking using sensor networks. It consists of a Hierarchical Hybrid Clustering Mechanism (HHCM) with a Prediction-based Algorithm for Destinationestimation (PAD). The proposed framework can track the destination of the object without prior information of the objects movement, while providing significant reduction in energy consumption. The costs of computation and communication are also reduced by collecting the most relevant information and discarding irrelevant information at the initial stages of communication. The contributions of this thesis are: Firstly, a novel Prediction-based Algorithm for Destination-estimation (PAD) has been presented, that predicts the final destination of the object and the path that particular object will take to that destination. The principles of origin destination (OD) estimation have been adopted to create a set of trajectories that a particular object could follow. These paths are made up of a number of mini-clusters, formed for tracking the object, combined together. PAD also contains a Multi-level Recovery Mechanism (MRM) that recovers tracking if the object is lost. MRM minimises the number of nodes involved in the recovery process by initiating the process at local level and then expanding to add more nodes till the object is recovered. Secondly, a network architecture called Hierarchical Hybrid Clustering Mechanism (HHCM) has been developed, that forms dynamic mini-clusters within and across static clusters to reduce the number of nodes involved in the tracking process and to distribute the initial computational tasks amoung a larger number of mini-cluster heads. Lastly, building upon the HHCM to create a novel multi-hierarchy aggregation and next-step prediction mechanism to gather the most relevant data about the movement of the tracked object and its next-step location, a Kalman-filter based approach for prediction of next state of an object in order to increase accuracy has been proposed. In addition, a dynamic sampling mechanism has been devised to collect the most relevant data. Extensive simulations were carried out and results were compared with the existing approaches to prove that HHCM and PAD make significant improvements in energy conservation. To the best of my knowledge the framework developed in unique and novel, which can predicts the destination of the moving object without any prior historic knowledge of the moving object