This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: A Novel Approach to Grid Sensor Networks Duminda A. Dewasurendra, Peter H. Bauer Dept. of Electrical Engineering University of Notre Dame Notre Dame IN 46556 USA email@example.com, firstname.lastname@example.org Abstract — A novel approach to information processing in grid sensor networks is presented. Based on the Fornasini-Marchesini (FM) model, this highly scalable method can implement any general linear system in a grid sensor network in a fully distributed manner. It can significantly reduce the communication overhead and enable local actuation in response to local events. Quantization effects on the stability of such distributed filters are briefly discussed with models for quantization nonlinearities. A real world implementation of this distributed processing method on a sensor network is also presented. Keywords : Grid sensor networks, Fornasini-Marchesini model, m-D systems, Distributed Filtering, Quantization nonlinearities I. INTRODUCTION Wireless sensor networks consisting of large numbers of resource- constrained embedded sensor nodes has recently become an emerging candidate for a multitude of distributed applications. Some of these applications require regularly placed nodes in a spatial grid, often sampling the sensors periodically over time. The research in  dis- cusses such networks for structural integrity monitoring. Agriculture and environmental monitoring includes other applications that often prefer a grid or mesh topology. Spatially distributed sensor lattices are also essential in certain surveillance, target location and tracking applications . Lattice sensor networks are discussed in detail in –, highlighting the network capacity, robustness to failures, and optimal routing schemes. Distributed information processing schemes are natural candidates for such networks with regularly placed nodes, yielding significant benefits in terms of scalability, reduced communication costs, energy savings and improved system lifetime. Furthermore, applications requiring local actuation in response to a local detection  are best supported by such distributed algorithms, yielding minimum response delays as compared to centralized schemes. A. Motivation and Goals In this paper, we present an approach to implement a general distributed application in a regularly placed grid sensor networks using a linear state-space model. It offers the following advantages for regularly-placed grid sensor networks. (a) General model : Any linear system can be implemented in the sensor network using this approach. (b) Highly scalable : This scheme is fully distributed and the com- munication and processing tasks are evenly distributed across all nodes. (c) Easily reconfigurable : All nodes run identical code. Hence, epidemic over the air programming schemes can be used to quickly reprogram thousands of nodes in the network for a different application....
View Full Document
- Spring '11
- Electrical Engineering