Energy-Aware Object Tracking Sensor Networks Yingqi Xu Department of Computer Science and Engineering Pennsylvania State University University Park, PA 16802 E-Mail: yixu@cse.psu.edu Advisor: Wang-Chien Lee Abstract and aggregation. These concepts are not simply juxtaposed, but fitting into each other to justify an integrated research topic. In order to fully realize the potential of sensor networks , energy awareness should be incorporated into every stage of the network design and operation. As a representative application of sensor networks, object tracking sensor network faces many system and design challenges in terms of energy management. By studying the role of semantic location in object tracking queries, data dissemination and sensing data storage, we propose a power-aware networking paradigm involving sensor nodes, networks protocols and application softwares for object tracking sensor networks. 1 2 Prior Research In order to realize energy-awareness in object tracking sensor networks, many issues from different aspects need to be addressed and integrated. These issues include: • data storage model, which defines the logical schema for data storage and distribution, • data dissemination model, which describes the mechanisms information (including sensing data and query) flows in the network, and • data model and query language, which provides the interface between users and the physical world. The above models are interrelated. For example, the distribution of the sensing data directly affects the efficiency and effectiveness of queries; on the other hand, the way users query the network plays important roles in defining the data schema. The Models should also satisfy the requirements of energy consumption, computational complexity, and other aspects of object tracking (e.g., latency and precision of detection). Many groups and researchers have been working on these topics in the context of general sensor networks. Three data storage models, namely Data-centric storage, Local storage and External storage are discussed in [3, 4]. They claim that using naming data in data storage and communication can significantly enhance the lifetime of sensor networks. Directed diffusion proposed in [2] is a userinitiated data dissemination protocol. By incorporating energy-aware techniques, such as data-centric dissemination, reinforcement-based adaptation, and innetwork data aggregation and caching, directed diffusion enables queries to extract the desired data with minimum number of transmissions, thus reducing the Introduction The rapid development of wireless technologies and computer-embedded devices has made it possible for people to monitor, control and interact with the physical world via sensor networks. While holding promise in a wide variety of applications, sensor networks are driven by extremely frugal battery resources, which necessitates the network design and operation to be done in an energy-aware manner. Object tracking sensor network, as a distinctive application of sensor networks, is widely used for military area intrusion detection and for wildlife animal monitoring. Because it involves a large amount of cooperation among sensors, object tracking sensor network provides significant research opportunities in terms of energy management. In order to maximize the lifetime of sensor networks, the system needs aggressive energy optimization techniques, ensuring that energy awareness is incorporated not only into individual sensor nodes but also into groups of cooperating nodes and into an entire sensor network. Based on the remarks, we plan to investigate the nature of the data flow in object tracking sensor networks, and focus on providing an energyefficient data management model for these networks. This involves query processing, network communication, sensor node cooperation, and data flow diffusion 1 power consumption. Two-tier Data Dissemination Model(TTDD) proposed by Ye et al. in [6] takes into account the mobility and scalability of sensor networks. As a source-initiated data dissemination model, TTDD handles the ”handoff” problem of mobile sinks in a power-efficient way by avoiding to flood the network with the location information of the moving sink. Prediction-based monitoring paradigm (PREMON) [1] investigates the problem of sensing data aggregations by exploring temporal and spatial relationships among sensor readings. The authors claim that sensors in close proximity are likely to have correlated readings. By predicting the future readings at a sensor based on given reading history and knowledge of surrounding sensors, PREMON reduces the number of transmissions at the cost of more receptions. 3 abstracting detection areas. With the abstraction of the physical world, the data model that describes the way data are stored, accessed, and operated is to be explored. After establishing logical schema for sensor data storage and a data model for applications, how the location information directs data flow inside the object tracking sensor network will be studied. The ultimate goal of our research is to build up a locationoriented data management model, which incorporates both user-initiated and sensor-initiated data diffusion and adapts to object behaviors and environmental changes. 4 Conclusion Through our study of data management and communication issues in object tracking sensor networks, we believe that semantic location may provide rich information for directing data dissemination and extracting data of sensors in an power-efficient way. This inspires us to propose a semantic location-based data model for object tracking sensor networks in order to achieve network-wide energy optimization. We envisage that this original idea may lead to new findings that significantly enhance the lifetime and performance of object tracking sensor networks. Proposed Research Many other related mechanisms exist, but limited space does not permit a thorough survey. To the best knowledge of the authors, no existing study addresses the research issues of data management in object tracking sensor networks. For example, due to the simplicity of the temporal and spatial prediction models, PREMON is more suitable for simple sensor networks monitoring than object tracking. We pointed out its deficiency and proposed a Dual Prediction approach for object tracking sensor networks in a previous work [5]. Furthermore, existing work in object tracking sensor networks uses IP-address, geometric location of sensor nodes, or description of tasks as keys for queries and data diffusion, but semantic locations1 have never been investigated as the predicates of queries. We believe incorporating semantic location information into data management model may become beneficial to energy consumption and network performance. As a next step to our previous study, we plan to investigate adaptive prediction models for object tracking sensor networks and how to intelligently and dynamically switch between various prediction models based on different movement patterns of mobile objects and application requirements. Moreover, to achieve power-aware prediction, we will identify and exploit the various performance-energy tradeoffs. Furthermore, we will study the role of the semantic location in object tracking sensor networks. For incorporating semantic location into sensor data management, an appropriate location model is necessary for References [1] S. Goel and T. Imielinski. Prediction-based monitoring in sensor networks: taking lessons from MPEG. ACM Computer Communication Review, 31(5), October 2001. [2] C. Intanagonwiwat, R. Govindan, and D. Estrin. Directed diffusion: a scalable and robust communication paradigm for sensor networks. In Mobile Computing and Networking, pages 56–67, 2000. [3] B. Krishnamachari, D. Estrin, and S. Wicker. Modelling data-centric routing in wireless sensor networks. [4] S. Ratnasamy, D. Estrin, R. Govindan, B. Karp, S. Shenker, L. Yin, and F. Yu. Data-centric storage in sensornets, 2002. [5] Y. Xu and W. C. Lee. On localized prediction for power efficient object tracking in sensor networks. In 1st International Workshop on Mobile Distributed Computing, May 2003. [6] F. Ye, H. Luo, J. Cheng, S. Lu, and L. Zhang. A twotier data dissemination model for large-scale wireless sensor networks. In Proceedings of the eighth annual international conference on Mobile computing and networking. ACM Press, 2002. 1 semantic locations is not the exact geometric location of individual sensor nodes, but a application specific area which covers a set of sensor nodes 2