Leveraging Single Board Computers for Anomaly Detection in the Smart Grid

Author USMA Department

Electrical Engineering and Computer Science

Document Type

Conference Proceeding

Publication Date

Fall 10-19-2017


Smart grid, Single board computer, Raspberry Pi, Phasor measurement units, Anomaly detection, Real-time systems, Computer architecture, Power grids


Smart Grid Technology is becoming an integral part of ensuring reliable and resilient operation of the power grid. The high sample rate and time synchronization of Phasor Measurement Units (PMUs) can provide enhanced situational awareness and more detailed information on power system dynamics as compared to traditional SCADA systems. A smart grid system must be able to detect alarm events (such as sudden voltage fluctuations or drops in current) in close to real-time. However, the communication network and bandwidth requirements to transfer large amounts of PMU data for realtime analysis is problematic. In this paper, we propose the use of a decentralized architecture for rapidly analyzing PMU data using single board computers to provide energy efficient monitoring locally in the power grid. This approach reduces communication requirements and enables real-time analysis. We present a novel anomaly detection scheme and test our approach on a real dataset of 1.4 million measurements derived from 8 PMUs from a 1000:1 scale emulation of a working power grid. Our results show that a single Raspberry Pi is sufficient to analyze data from multiple PMUs at a rate suitable for real-time analysis.

First Page


Last Page



Funding for the SBC hardware and algorithmic development was provided by the Office of Naval Research. Funding for the initial algorithm development was provided by U.S. Army Armament Research, Development and Engineering Center (ARDEC).

Conference Name

2017 IEEE 8th Annual Ubiquitous Computing, Electronics and Mobile Communication Conference (UEMCON)

Conference Location

New York, NY

Conference Dates