EMKOS: Design and Implementation of Microkernel OS for Edge Computing

Abstract:
The operation of Internet of things (IoT) devices generates a large amount of data, most of which is uploaded to the central cloud for computational processing, which increases the latency and power consumption of devices and systems. Existing research focuses on unloading tasks to the network edge for execution to reduce latency and energy consumption, but in real-world applications, there is a need to customize and optimize the IoT operating system to meet the requirements of fast unloading of computational tasks. To this end, an Edge Microkernel Operating System (EMKOS) is designed and developed to achieve fast unloading of computational tasks on resource-constrained IoT devices by using Dynamic Linking and Dynamic Loading (DL²) technology. The EMKOS system uses DL² technology to achieve fast allocation and loading of computational unload tasks on resource-constrained IoT devices, reducing computational latency and system power consumption. Meanwhile, an edge computing-based computational unloading simulation environment is built to simulate the computational unloading of three types of application scenarios, and the experimental analysis of the three types of application scenarios in real use is conducted in combination with EMKOS. The experimental results prove that the system can be widely applied to application areas such as smart home, Internet of Things, Internet of Vehicle, and smart campus, improving system efficiency and reducing system energy consumption.