University of Houston

We can learn a lot about the activities performed in a computing system and the health of a system by measuring and analyzing its energy consumption. Analysis of energy consumption with the goal of understanding the computing system in an uncontrolled environment is an open research area. The main challenge is to disaggregate patterns from a large amount of discrete time series data coming from energy measurements.

The purpose of this study is to understand how the system workloads and activities can be inferred from energy usage of computing systems and networks. The specific objectives of this study include: (a) analysis of the relationship between system workloads and energy used by the systems, (b) disaggregation and classification of the system workloads only from energy consumption, (c) modeling and predicting whole network's energy consumption based on the features recognized from partial energy readings, (d) improving the energy efficiency of computing systems by helping identify sources of waste.