           !" ## 21            Elham Ahmadi Department of Computer Engineering, Islamic Azad University,South Tehran Branch, Tehran, Iran Masoud Sabaei Department of Computer Engineering and Information Technology, Amirkabir University, Tehran, Iran Mohamad Hosain Ahmadi Department of Electronic Engineering, Smart Company Tehran, Iran  These days mobile target tracking is considered as one of the important applications of wireless sensor networks. In this regard, the clustering structure is one of the most applicable network structures. In this paper, we suggested a new method for target tracking that makes adaptation with target mobility model. This method utilizes two tools to create adaptability which are changing the size and shape of clusters according to target mobility model. Also by combining static and dynamic clustering a semi dynamic clustering structure has been developed to implement these tools. Simulation results show that our suggested method decreases both energy consumption by decreasing clusters size when the target moves uniformly, and tracking error by changing the size and the shape of clusters according to target mobility when the target moves unpredictably.   Semi!Dynamic clustering, target tracking, tracking error.  Wireless sensor networks, target tracking, tracking error, target mobility model.   Sensor network (WSN) is composed of a large number of sensor nodes and deployed either inside the phenomenon or very close to it. Wireless sensor networks are expected serve as a key infrastructure for a broad range of applications including precision agriculture, surveillance highway systems, emergent disaster response and recovery. One of the important application issues for sensor networks is utilized to track mobile object [1]. Researches about target tracking with continuous monitoring mechanism can be divided into three categories: structure!less scheme, tree!based scheme, and cluster!based scheme. In structure!less scheme, when a node detects an object within its range, it broadcasts a ‘TargetDetected’ message. This message contains the location of the sensor node and the distance to the target. All nodes that hear this message store its data in their local memory. When a node that has detected the target hears two other ‘TargetDetected’ messages from two other nodes, it performs triangulation on the three coordinates to calculate the location of the target [2]. Tree!based scheme uses a hierarchy to connect the sensors. In this scheme, when a target shows up for the first time, an initial convey tree is constructed and the root collects data from nodes surrounding the target, and processes the data. When the target moves, the tree’s membership is changed and its structure is reconfigured if necessary [3]. To facilitate collaborative data processing in target tracking centric sensor networks, the cluster architecture is usually used in which sensors are organized into clusters, and each cluster consists of a cluster head (CH) and several neighboring sensors (members). In the static clustering architecture, clusters are formed statically at the time of network deployment. The attributes of each cluster, such as the size of a cluster, the area it covers, and the members it possesses, are static. In spite of its simplicity, this structure suffers from several drawbacks; for example, it can’t adapt itself with the average error of tracking because of the fixed number of nodes in each cluster. Also if a CH dies of power depletion, all the sensors in the cluster render useless [4]. In dynamic clustering structure, when target moves, the nodes that recognize the target gradually form dynamic clusters. This method is more accurate than static clustering structure but has higher calculation and communication overhead [4, 5]. In summary, the clustering architecture is more applicable for tracking applications and it has less overhead rather than other structures [4]. Regarding these cases, we decided to suggest a new clustering structure that, in contrast with static clustering, can change clusters size according to tracking error value and have less calculation and communication overhead rather than dynamic clustering. In addition, few researches of target tracking are concerned with the target mobility model type. Therefore, we decided to suggest a new tracking protocol that is adaptable with the target mobility model. The remaining of this paper is organized as follows. In section 2, we first review some related works. In section 3, our proposed target tracking method is introduced and described. In section 4, we compare energy consumption in our structure with other clustering structures (static and dynamic clustering). In section 5, we analyze and evaluate this approach throughout simulation. Finally, we conclude the paper in section 6 with a summary.    Paper [6] has proposed a density based clustering for node management that is a decentralized algorithm having the topology control information in each sensor node. The authors in [7] have proposed an object tracking strategy named multi level object tracking (MLOT) which is based on multi!level architecture for efficient object tracking and real!time recovery of missing objects by mining the movement log in sensor networks. In [8] a distributed protocol for target tracking based on static clustering has been developed that predicts the target’s next location using a linear predictor. Also an efficient dynamic clustering algorithm for object tracking in [9] has been introduced that proposes a new cluster head election algorithm