Green Access Control System

Abstract: The number of fixed-installation resource-constrained devices, such as Internet of Things devices, has been growing rapidly in the past years. One challenge of accommodating these large numbers of connected devices is the need to power them all. In this thesis, we investigate ways to reduce the power consumption of the door controller in an existing physical access control system, without compromising on system functionality or latency. Existing power management solutions of other resource-constrained devices were evaluated and suggested. Protocols and hardware components present in the system were researched, with focus on how power consumption and latencies can be reduced. Power measurement tests were performed on Linux power management systems CPUFreq and CPUIdle to evaluate their impact, as well as when suspending the CPU. Our results show that CPUFreq and CPUIdle are a simple way to reduce overall power consumption without compromising on system latency, and that over 25% of total power can be saved by suspending the CPU when the device is not in use. These results suggest that the greatest power savings are found when suspending the door controller CPU, and that system adjustments must be made to accommodate a suspended CPU. With this work, we hope that this type of device could adopt a battery-based power solution, reducing upfront installation costs.