PowerSAX: Fast motif matching in distributed power meter data using symbolic representations

Distributed power meters (also termed smart plugs) are embedded systems that measure the electric power consumption of individual appliances at a fine temporal resolution. They enable a wide range of novel smart services, e.g., accurately forecasting power consumption or making recommendations how to save energy. However, distributed power metering combines high sampling rates with a potentially large number of monitored outlets. A torrent of power readings may thus be generated, incurring high bandwidth requirements for their transmission and a significant computational power demand for their processing. In this paper, we present a concept for the efficient local storage and processing of power consumption data called PowerSAX. Instead of operating on raw sensor readings, PowerSAX converts consumption data into their symbolic representations and thus mitigates their storage requirement. Subsequently, it enables embedded systems to recognize relevant patterns (motifs) in the symbolic representations of collected data. By only transmitting a message when a known motif is encountered in the sensor data, PowerSAX can significantly reduce an application's bandwidth requirements. We evaluate PowerSAX using real-world power consumption data and show to which extent smart plugs can make predictions of an appliance's future power demand.

[1]  Gregory D. Abowd,et al.  At the Flick of a Switch: Detecting and Classifying Unique Electrical Events on the Residential Power Line (Nominated for the Best Paper Award) , 2007, UbiComp.

[2]  Shwetak N. Patel,et al.  ElectriSense: single-point sensing using EMI for electrical event detection and classification in the home , 2010, UbiComp.

[3]  G. W. Hart,et al.  Nonintrusive appliance load monitoring , 1992, Proc. IEEE.

[4]  Suhuai Luo,et al.  An Approach of Household Power Appliance Monitoring Based on Machine Learning , 2012, 2012 Fifth International Conference on Intelligent Computation Technology and Automation.

[5]  Eamonn J. Keogh,et al.  HOT SAX: efficiently finding the most unusual time series subsequence , 2005, Fifth IEEE International Conference on Data Mining (ICDM'05).

[6]  Eamonn J. Keogh,et al.  A symbolic representation of time series, with implications for streaming algorithms , 2003, DMKD '03.

[7]  Shahram Latifi,et al.  A survey on data compression in wireless sensor networks , 2005, International Conference on Information Technology: Coding and Computing (ITCC'05) - Volume II.

[8]  Marc Boullé,et al.  SAXO: An optimized data-driven symbolic representation of time series , 2013, The 2013 International Joint Conference on Neural Networks (IJCNN).

[9]  Ralf Steinmetz,et al.  On the accuracy of appliance identification based on distributed load metering data , 2012, 2012 Sustainable Internet and ICT for Sustainability (SustainIT).

[10]  Andreas Reinhardt,et al.  Predicting the Power Consumption of Electric Appliances through Time Series Pattern Matching , 2013, BuildSys@SenSys.

[11]  Dorin Comaniciu,et al.  Mean Shift: A Robust Approach Toward Feature Space Analysis , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[12]  Dina Q. Goldin,et al.  On Similarity Queries for Time-Series Data: Constraint Specification and Implementation , 1995, CP.

[13]  Kay Römer.  Discovery of Frequent Distributed Event Patterns in Sensor Networks , 2008, EWSN.

[14]  Jeffrey D. Scargle,et al.  Astronomical Time Series Analysis , 1997 .

[15]  Karl Aberer,et al.  Symbolic representation of smart meter data , 2013, EDBT '13.

[16]  Ralf Steinmetz,et al.  Scoresheet-based event relevance determination for energy efficiency in wireless sensor networks , 2011, 2011 IEEE 36th Conference on Local Computer Networks.

[17]  Jessica Lin,et al.  Finding Motifs in Time Series , 2002, KDD 2002.

[18]  Steven B. Leeb,et al.  Power signature analysis , 2003 .