Granger causality analysis on IP traffic and circuit-level energy monitoring

  • Authors:

  • Younghun Kim;Rahul Balani;Han Zhao;Mani B. Srivastava

  • Affiliations:

  • University of California, Los Angeles;University of California, Los Angeles;University of California, Los Angeles;University of California, Los Angeles

  • Venue:
  • Proceedings of the 2nd ACM Workshop on Embedded Sensing Systems for Energy-Efficiency in Building
  • Year:
  • 2010

Quantified Score

Hi-index 0.00

Visualization

Abstract

Device-level energy monitoring has been increasingly proposed to understand inefficient energy use and design systematic processes for efficient building operation. Its sole use, however, is not sufficient to provide actionable information unless we understand the causes and context of energy use. Fundamentally, energy consumption in a building is due to occupants' various activities. Understanding the causal relationship between occupants and their energy use is thus the key to an efficient building operation. This usually involves fine-grained sensing through intensive instrumentation of individual power outlets and/or extensive user studies that either increase the system cost or become too intrusive. Instead, we advocate that circuit branch level energy monitoring combined with statistical Granger causality analysis is adequate to automatically understand the causal relationship. We monitor energy consumption of various zones in an office using a circuit level power monitor. IP traffic from users' PCs, obtained from a local firewall, is used to relate occupants with their energy use in each micro zone. The output is expressed in the form of causality graphs that illustrate how each individual influences energy use in different zones. We discuss the effectiveness and limitations of this causal analysis in capturing energy use patterns of the occupants in a lab environment.