A sensor network for predicting and maintaining occupant comfort

This paper describes a sensor network that has been designed and deployed to monitor the environmental conditions within a large multi-use building. The conditions measured across this network include temperature, humidity, light levels, ambient sound level, room occupancy and an informal survey of occupant comfort level. The nodes were designed to be low cost with the intention that a significant number of them could be deployed around large buildings. In the initial study described in this paper, a network comprising 17 prototypes nodes has been deployed in a multi-use university building. The sensor network is Ethernet based and uses the building's existing infrastructure for communication with a central server which stores the recorded data. Preliminary analysis of the data collected from this sensor network over a period of 10 months combined with data from a local meteorological station and data from the building's fiscal energy meters is presented. The overall goal of the project is to use the data from the network to find more energy efficient methods of heating the building while maintaining occupant comfort. In particular, it is anticipated that the data collected from the network will identify parameters, other than temperature, which are indicative of occupant comfort. To determine this, the current work is focussed on building a classifier to predict occupant comfort level, given the data measured by the sensor nodes combined with data from a local weather station and the building's smart energy meters. Analysis of the identified model should indicate which variables most influence occupant comfort. The project also aims to produce a dataset which may be used for future research and aid design of sensor networks.