Skip to Main Content
Since the inception of sensor networks, it is recognized that the foremost application of such a network is on monitoring environmental activities. Due to the extreme resource constraints, a critical feature of such a network is to conserve energy to extend the lifetime of the sensors. A variety of energy aware algorithms have been developed for these networks by aggregating or filtering of data being transmitted. Energy can also be saved by selecting a subgroup of sensors to activate each time. Uncertainties are introduced due to the inaccuracy in sensor readings, noise or failure of sensors, or the omissions of data transmission following those energy aware algorithms. On the other hand, there usually are uncertainties inherited in the application where these sensor networks are built. Its usefulness depends on whether we can find a strategy to reduce uncertainties and at the same time conserve energy in such a system. We discuss the application of sensor networks in the problem of syndromic surveillance. The aggregated action of the sensors would provide early evidence for screening and identification of outbreaks of diseases or bio-agents. We study the effects of using the Bayesian methods to reduce uncertainties in sensor networks for medical decision-making. Experimental results obtained using operational data are used to verify our assumptions. A lot of study has been done on applying sensor networks to tasks such as tracking of a moving object, which is mapping an estimation task over a network of sensors. This project is the first case to map a medical decision, which is essentially an uncertainty refinement problem to sensor networks.