UC Irvine

We propose a new system which can recognize and classify the different operational states of individual devices in home entertainment systems based on their power consumption models with the goal of improving energy management systems, providing information for energy saving, and reducing energy wasting. Modern advanced power strips can provide energy saving solutions by applying different kinds of sensors to monitor the user activities and make the decision to cut off the power source of the controlled outlet. However, normally this kind of solution needs a long trigger time and has less accuracy due to the differences in user behavior. Our system is able to effectively identify a device’s class and its operational states to provide a more efficient solution for home entertainment systems’ energy saving. We use LabVIEW and power analyzer to collect real time power consumption data for the different states of individual devices every second. Tracking and understanding this data will let us establish the dynamic models of the operational states in home entertainment systems. We devise recognition techniques based on those dynamic models by characterizing these devices’ signatures and their power consumption distributions to build an algorithm to recognize the operational states of different devices. We designed a prototype integrated with our algorithm and models that demonstrated up to 60% wasted energy could be saved. This system would enable home entertainment systems to become more energy efficient and further reduce energy waste by combining user behavior tracking and prediction.