Demo Abstract: Scalable Load Disaggregation System Using Distributed Electrical Signature Detection ∗

We design and demonstrate a load disaggregation system that selectively uses high-frequency data from an add-on battery-powered mote hosting a current transducer (CT) sensor to complement low-frequency data from an infrastructure power meter. Our system learns, detects, and combines coarse electrical signatures from the infrastructure meter and finer signatures from the CT sensor to infer energy usage of targeted end-loads in real time. We demonstrate a proof-ofconcept prototype of our system under a set of controlled experiments with various electrical loads.