Sensor Validation and Fusion with Distributed ‘smart Dust’ Motes for Monitoring and Enabling Efficient Energy Use

MEMS (microelectronic mechanical systems) sensors make a rich design space of distributed networked sensors viable. They can be deeply embedded in the physical world and spread throughout our environment like “smart dust”. Today, networked sensors called Smart Dust motes can be constructed using commercial components on the scale of a square inch in size and a fraction of a watt in power. Highdensity distributed networked sensors have recently been targeted for use in research devoted to the efficient use of energy. Such networks require a large number of sensors for control at different levels. However, in reality, sensor information is always corrupted to some degree by noise and degradation, which vary with operating conditions, environmental conditions, and other factors. To overcome these shortcomings, sensor validation is needed to assess the integrity of the sensor information and adjust or correct as appropriate. Sensor fusion of both disparate and redundant (physical and functional) sensors is essential for control and to achieve high sensor data fidelity. In this paper we have isolated a specific domain within the built environment, and present an influence diagram model by which to answer the decisions concerning how to most efficiently condition that space throughout the day. Key issues include the aggregation of heterogeneous information, management of uncertainty at various levels, appropriateness assessment of current validation and fusion algorithms, temporal changes and decision-making strategies. 1. Distributed MEMS Sensors – Smart Dust The goal of the Smart Dust project is to build a selfcontained, millimeter-scale sensing and communication platform for a massively distributed sensor network [16,24]. These devices will ultimately be around the size of a grain of sand and will contain sensors, computational ability, bi-directional wireless communications, and a power supply, while being inexpensive enough to deploy by the hundreds. There exist endless possibilities for the applications of these devices. The wireless sensing capabilities of Smart Dust, combined with its microscopic size give everything in the physical world the potential to be “smart.” As the cost of the devices decreases, and the number of sensors collated into a network increases, the need for sensor fusion and validation techniques becomes increasingly necessary. Similarly, as the real-world applications of these sensor webs become increasingly sophisticated, the integrity of the information conveyed within the web, and the decisions that can be executed merit increasing levels of attention. Before turning to the details of the research questions afforded by Smart Dust applications, we review the basic architecture and design of the macroscale motes available today. From: AAAI Technical Report SS-02-03. Compilation copyright © 2002, AAAI (www.aaai.org). All rights reserved.