Falcon: An ambient temperature aware thermal control policy for IoT gateways

Abstract

With the blooming of the Internet of Things (IoT), the demand for smart gateways that can offload the computation and communication tasks of IoT equipment is fast growing. Gateways are the bridge between the connected devices and the Internet, and their reliability is very important. According to literature, the reliability of electronic hardware components is closely related to their operating temperature. To avoid system performance loss under the thermal constraints, system designers rely on fans to cool the system. The problem is that the over cooling will produce noise and waste energy. In this paper, we propose Falcon, an ambient temperature aware thermal control policy for smart gateways. Falcon leverages a runtime thermal prediction model and takes the ambient temperature into the consideration to tune the fan speed proactively. Falcon can save 4.85% total system power and reduce noise in 34% of the time on average. In addition, the two environment adaption mechanisms applied in Falcon help us achieve the same thermal control ability in different ambient temperature environment, especially in high ambient temperature cases.