Infrastructure-Mediated Sensing Based Home Appliances Monitoring System Using the EMI Characteristics

An automatic monitoring system for home appliances using infrastructure-mediated sensing technology is presented. A practical solution for identifying the working states of electronic devices and determining the consumption of each device in a household is proposed. Initially, the components of Electromagnetic interference (EMI) noises in a residential power line are analyzed, and then the basic theory of Switched mode power supplies (SMPS) is discussed. Finally, a set of practical approaches has been proposed to detect and classify the electrical events, including a time-frequency transformation algorithm, power spectrum vector chasing, Gaussian function fitting and supervised pattern recognition, etc. The monitoring system has deployed into real houses and it successfully classifies SMPS devices with accuracy from 92.6% to 99.2% for individual devices and 88.0% to 95.0% for multiple devices.