TRUSTWORTHY OPTIMIZED CLUSTERING BASED TARGET DETECTION AND TRACKING FOR WIRELESS SENSOR NETWORK

C JEHAN AND D SHALINI PUNITHAVATHANI: TRUSTWORTHY OPTIMIZED CLUSTERING BASED TARGET DETECTION AND TRACKING FOR WIRELESS SENSOR NETWORK DOI: 10.21917/ijct.2016.0195 TRUSTWORTHY OPTIMIZED CLUSTERING BASED TARGET DETECTION AND TRACKING FOR WIRELESS SENSOR NETWORK C. Jehan1 and D. Shalini Punithavathani2 1 Department of Computer Science and Engineering, Tamizhan College of Engineering and Technology, India E-mail: , 2 Government College of Engineering, Tirunelveli, India E-mail: , Clustering based Target Tracking scheme has three steps. In the first step, cluster identification and cluster head selection by means of Oppositional Gravitational search optimization (OGSA) with kernel Fuzzy C-means clustering (KFCM) algorithm. Initially, oppositional gravitational search algorithm (OGSA) is utilized to optimize the initial clustering center. Afterwards, the KFCM algorithm is availed to guide the classification and the cluster generation process. The second step, the Cluster Head Data Transmission step deals with the transmission of data from all the members of the cluster to the cluster head within the cluster. The cluster head aggregates all the received data. The third step is the Base Station Tracking step, in which all the cluster heads transmit the aggregated data to the base station. This tracking step detects the target. The paper proceeds as follows. The section 2 describes the related works and the section 3 explains the network model, WSN formation and GSA algorithm. The section 4 explains the proposed scheme and the section 5 presents the performance evaluation of Adaptive Dynamic Clustering Target Tracking scheme and summarizes the result. Finally, section 6 concludes the paper. Abstract In this paper, an efficient approach is proposed to address the problem of target tracking in wireless sensor network (WSN). The problem being tackled here uses adaptive dynamic clustering scheme for tracking the target. It is a specific problem in object tracking. The proposed adaptive dynamic clustering target tracking scheme uses three steps for target tracking. The first step deals with the identification of clusters and cluster heads using OGSAFCM. Here, kernel fuzzy c-means (KFCM) and gravitational search algorithm (GSA) are combined to create clusters. At first, oppositional gravitational search algorithm (OGSA) is used to optimize the initial clustering center and then the KFCM algorithm is availed to guide the classification and the cluster formation process. In the OGSA, the concept of the opposition based population initialization in the basic GSA to improve the convergence profile. The identified clusters are changed dynamically. The second step deals with the data transmission to the cluster heads. The third step deals with the transmission of aggregated data to the base station as well as the detection of target. From the experimental results, the proposed scheme efficiently and efficiently identifies the target. As a result the tracking error is minimized. Keywords: Clustering, Dynamic, Gravitational Search Target Tracking, Static, Oppositional, 2. LITERATURE REVIEW 1. INTRODUCTION A number of approaches have been developed meta-heuristic algorithms to solve various problems in WSNs. Among them, a small number of approaches are discussed in this section. In [8], the authors propose a location tracking methodology based on radio waves. These employ received signal strength to calculate the location of an object. Their technique basically speaks about selecting a set of points and then based on the RF connectivity between these points; the transmitting sensors are placed only on a subset of 15 these points. The sensors have a limited range of transmission and the observer would receive unique ID packets anywhere in this region. In [9], the author identifies a cluster head which is responsible for implementing the algorithm. In order to minimize traffic and conserve energy, a notification is sent by a sensor to the cluster-head whenever the object is tracked which then queries a subset of sensors to gather more detailed target information. These are intelligent queries based on the cluster-head generating a probability table for each grid point and then subsequent localization if a target is detected by one or more sensors. The aim of the algorithm in [7] is that every node that has at least one spatial neighbour that is a Delivery-Zone node that will forward or locally broadcast the Mobicast packet. So, all delivery zone nodes will receive the corresponding packet. This simple rule leads an ‘as-soon-as-possible’ style Mobicast protocol that One of the important application of wireless sensor network is target tracking [5] [6] [10]. The main focus of target tracking in WSN is localization and object tracking. For a reliable target tracking system, good tracking quality and energy efficiency are the two very important requirements [1] [3]. In order to save energy as well as to provide good trade off between energy efficiency and tracking accuracy, cluster-based tracking has been preferred for target detection. In cluster tracking system, the nodes are grouped into clusters. The cluster based target tracking can be further divided into two approaches, static clustering and dynamic clustering [2] [12] [14]. In static clustering, clusters are formed statically at the time of network deployment and the clusters are formed before the intruders enter into the network; where as in dynamic clustering clusters are formed dynamically and the clusters are formed by using the signal strength from the target [4] [15] [16] and [21]. In this paper, a fully decentralized Adaptive Dynamic Clustering based Target Tracking scheme has been proposed for single target tracking. The cluster heads are chosen based on the energy levels of the nodes in the cluster. This data is transmitted from all the members of the cluster to the cluster head. Retransmission of data is done on timely manner. The proposed 1326 ISSN: 2229-6948(ONLINE) ICTACT JOURNAL ON COMMUNICATION TECHNOLOGY, JUNE 2016, VOLUME: 07, ISSUE: 02 exhibits a high average slack time which is not desirable. The author present self-organized distributed target tracking techniques with prediction based on Pheromones, Bayesian, and Extended Kalman Filter techniques [18], [19]. The implementation and testing of a real distributed sensor network collaborative tracking algorithm in a military context is described in [20]. Sandy Mahfouz et al. [28] described a technique for target tracking in wireless sensor networks. This approach combined machine learning with a Kalman filter to assess instantaneous positions of a moving target. The target's accelerations, along with information from the network, are utilized to achieve an accurate estimation of its position. In [29], they also introduced two major contributions to the wireless sensor network (WSN) society. The first one consists of modeling the relationship between the (...truncated)