Modelling Energy Consumption using Thermal Imaging
Name
Hashim Hashimov
Abstract
We contribute a novel sensing solution that can estimate the energy consumption of applications running in personal devices, such as smart and wearable devices, without requiring the instrumentation of the device itself. Energy estimation of applications running in these devices has become a complex challenge. Indeed, personal devices no longer have a detachable battery, such that their design can be further optimized to adjust better to the everyday activities of users. This in turn, it makes difficult to profile the energy consumption of software developed for these devices with typical solutions, like the Monsoon power meter. Our solution uses thermal imaging to derive energy consumption measurements while the applications are running in the device. We develop a functional prototype to demonstrate the feasibility of our solution. Through rigorous benchmarks that take into consideration different applications, we found that our solution can produce estimations of energy consumption that are comparable to existing hardware solutions. We also found that these estimations also capture differences between different types of applications. In addition, we also found that it is possible to use our solution to estimate the energy consumption of applications running in IoT devices. We also developed a regression model that predicts the estimated power from thermal imaging.
Graduation Thesis language
English
Graduation Thesis type
Master - Computer Science
Supervisor(s)
Huber Flores
Defence year
2022