Fine-grained monitoring

Existing electricity monitoring solutions often do not provide fine-grained information for the users (e.g., in the case of smart meters) or are too expensive for large-scale deployment (e.g., in the case of smart power plugs and CT-based electric sub-metering systems). Many of them (e.g., smart meters and CT-based electric sub-metering systems) also require professional electricians to install. There is a lack of user-friendly and low-cost system solutions that can help users understand their electricity usage with sufficient accuracy.


The central challenge is to achieve high estimation accuracy for fine-grained electricity usage information with low-cost sensing devices while incurring minimum installation / maintainance efforts from the end users. We also need to devise different solutions to suit the needs of different end users.

Fine-grained monitoring

Our approaches

We have invented multiple complementary solutions to provide end users with fine-grained electricity usage information in low-cost and easy-to-install ways (see the table below).

Approach Goal Device(s) needed Installation
Supero Estimate the states for home appliances with distinguishable physical signals, like light and sound. Multiple indirect sensors and an in-situ smart meter Distribute the sensors across the space to be monitored
Harmonic-Sense Estimate the states for home appliances with distinguishable harmonic patterns in current waveform A single battery-powered mote hosting a current transducer and an in-situ smart meter Install the sensor mote onto the main feeder of the home / office
Noninvasive Sub-metering Estimate the current readings of each branch in an electricity distribution panel An Anisotropic Magnetoresistive (AMR) sensor array and an in-situ smart meter Install the AMR sensor array on the electricity distribution panel

Our results

While our solution relies on low-cost sensors and allows quick deployment by non-professional users, we can still achieve high estimation accuracy. In particular, when tested extensively in real home environments, both Supero and Harmonic-Sense can estimate the energy consumption of individual appliances with less than 10% of error rate. Our noninvasive sub-metering solution based on AMR sensor arrays can estimate the per-branch current readings with less than 10% of error rate, even for branches carrying small loads (e.g., a 30W fan). See more details in the following two demo videos:

Load Disaggregation Video

Load Disaggregation Demonstration
presented at BuildSys 2012
Electric Sub-metering Video

AMR Electric Submetering Demo
presented at SenSys 2012


  • Dennis E. Phillips, Rui Tan, Mohammad-Mahdi Moazzami, Guoliang Xing, Jinzhu Chen, David K. Y. Yau
    Supero: A Sensor System for Unsupervised Residential Power Usage Monitoring
    IEEE International Conference on Pervasive Computing and Communications (PerCom’13)
    Best paper candidate
    San Diego, CA, USA, March 2013
  • Deokwoo Jung, Hoang Hai Nguyen, Sreejaya Viswanathan, Binbin Chen and David K.Y. Yau
    Demo: Scalable Load Disaggregation System Using Distributed Electrical Signature Detection
    In Proceedings of the 4th ACM Workshop On Embedded Systems For Energy-Efficiency In Buildings (BuildSys’12)
    Toronto, Canada, November 2012 [pdf]
  • Sreejaya Viswanathan, Binbin Chen, Hoang Hai Nguyen, Jerry T. Chiang, Deokwoo Jung, and David K. Y. Yau
    Demo: Using Anisotropic Magnetoresistive (AMR) Sensor Arrays for Electric Sub-metering
    In Proceedings of the 10th ACM Conference on Embedded Networked Sensor Systems (SenSys’12)
    Toronto, Canada, November 2012 [pdf]