عنوان مقاله
ردیابی و پیشگویی اهداف متحرک در شبکه های حسگر سلسله مراتبی
فهرست مطالب
چکیده
مقدمه
استراتژی پیشگویی سلسله مراتبی
نتایج شبیه سازی و تحلیل
نتیجه گیری
بخشی از مقاله
پیکره بندی مجدد تحت شکست های CH
به محض شکست CH برای دوره زمانی خاص، CH های نزدیک به این توافق می رسند که CH نظیر وجود ندارد. به همین منظور برای محاسبه تقسیم خوشه جدید، دور جدیدی از مذاکره مجدداً از سر گرفته می شود. NN هایی که CH شان را از دست داده اند توسطCH های جدید جذب خواهند شد. بنابراین پیکره بندی مجدد شبکه خاتمه می یابد.
کلمات کلیدی:
Tracking and Predicting Moving Targets in Hierarchical Sensor Networks Zhibo Wang, Hongbin Li, Xingfa Shen, Xice Sun, Zhi Wang* Abstract-Target tracking is an important application of newly developed Wireless Sensor Networks (WSN). Much work has been done on this topic using a plane network architecture. We propose a scheme, namely Hierarchical Prediction Strategy (HPS), for target prediction in hierarchical sensor networks. The network is divided into clusters, which are composed of one cluster-head and many normal nodes, by Voronoi division. For an existing target, cluster-heads only selectively activate nearby sensor nodes to perform tracking. Moreover, Recursive Least Square technique is used to predict the target trajectory and help activate next-round sensor nodes. Extended simulations show the properties of the proposed network architecture and the efficiency of the prediction scheme. Index Terms-Wireless Sensor Network, Target Tracking, Prediction. I. INTRODUCTION With the fast development of distributed Wireless Sensor Networks (WSN), the scenarios of field surveillance and target tracking by small and populated local sensor nodes become possible. The sensor nodes, integrated with sensing, data processing and wireless communication, are utilized to play significant roles in environment monitoring, traffic control, precise agriculture and battlefield surveillance [1]-[4]. Target tracking is an important application of WSN. However, large-scale wireless networks frequently suffer from packet losses, communication delays and energy limitations etc. How to coordinate a large-scale network to efficiently track a moving target while conserve network resources, namely energy and bandwidth, is a great challenge. A main feature, introduced by tracking in large-scale networks, is only those sensors near target can detect the target and perform sensing. Based on this fact, making all nodes working is not the optimal choice. Our efforts in this research mainly focus on the management of large-scale networks and the effectiveness of target tracking. By managing sensor nodes behaviors, only a small amount of nodes near the target are activated, thus saving energy without losing target. Moreover, activated cluster can perform target prediction and nodes preactivation, which enable high-quality tracking. Most recent work focused on totally distributed homogeneous sensor models and dynamic cluster coordination. As a matter of fact, energy resources, computation capacity and Zhibo Wang, Hongbin Li, Xice Sun and Zhi Wang are with the State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou, China. Xingfa Shen is with the Institute of Computer Application Technology, Hangzhou Dianzi University, Hangzhou, China. *Corresponding Author. (e-mail: wangzhi@iipc.zju.edu.cn) bandwidth are restricted in WSN. Such networks, managed by dynamic cluster coordination, are prone to suffer from endless message exchanges among sensor nodes, which drain off network energy quickly. In hierarchical network, however, powerful nodes, which can optimize local resources and efficiently coordinate local network, provide an alternative. In this paper, we propose a scheme named Hierarchical Prediction Strategy (HPS) for target tracking in sensor networks. The two-tier hierarchy divides sensor nodes as Cluster-Heads (CHs) and Normal Nodes (NNs). We assume that CHs can communicate with each other and send commands/inquiries to NNs, while NNs can transmit observations to CHs but do not talk to other NNs. By constraining NNs' communication and reducing network congestion, wireless bandwidth is saved so that fusion packets formed on CHs can be delivered with better guarantee. Further, we explore the property of the proposed tracking and predicting algorithm in the hierarchical network. Simulations show that our method efficiently coordinates large-scale hierarchical network and achieves a nice tradeoff between energy consumption and tracking quality. The outline of this paper is as follows. Section 2 briefly presents a survey of related work. In Section 3 we propose a scheme for target tracking in sensor network. Then Section 4 presents and analyzes the simulation results concerning our proposed scheme. Finally, Section 5 states our conclusions and future work. II. RELATED WORK The first cluster-based routing approach for dynamic network is introduced in [5]. Composed of a group of neighboring nodes, a cluster manages the whole detection/tracking process, with the cluster head collecting joint observations from the pertaining sensor nodes and eventually producing a final fusion results. Moving target tracking based on WSN has been an attractive topic in recent years. The information driven sensor querying (IDSQ) mechanism for sensor collaboration in ad hoc sensor networks is proposed in [6]. In IDSQ, at any time, there is a dynamic leader deciding which sensors should be selectively activated in order to obtain the best information about the target. In [7], the authors use cellular automata models to study tracking performance and network data in a unified framework. [8] describe self-organized, distributed target tracking techniques with prediction based on Pheromones, Bayesian, and Extended Kalman Filter. In [9], the authors assume a scenario of totally distributed WSN and the performance of four strategies, namely Naive Activation, Random Activation, 1169 Authorized licensed use limited to: Amirkabir Univ of Tech Trial user. Downloaded on July 01,2010 at 08:23:31 UTC from IEEE Xplore. Restrictions apply.
