Developing a power measurement framework for cyber defense

Energy Delivery Systems (EDS) have become smarter by incorporating chips and data communication capabilities. As a result, they have become more vulnerable to cyber-attacks as well. As part of a systems health monitoring approach, we investigate whether cyber-events targeting EDS can be detected by monitoring component-level data such as temperature, voltage, power, and process indicative variables---collectively referred to as component health indicators. We report our experiences with developing a measurement framework for power consumption in different EDS components such as Cabinet Distribution Units (CDU), Power Distribution Units (PDU), and standard enterprise desktops. Our plan for gathering and analyzing power consumption data involves establishing a baseline execution profile and then capturing the effect of perturbations in the state from injecting various malware. As a contribution, we report on initial experiments with power measurement techniques and outline future work for evaluating components under normal and anomalous operating regimes.