Light weight serverless computing at fog nodes for internet of things systems

Abstract

Internet of things (IoT) systems collect large size of data from huge numbers of sensors. A wide rage of IoT systems relies on cloud resources to process and analyze the collected data. However, passing large amount of data to the cloud affects the overall performance and cannot support real-time requirements. Serverless computing is a promising technique that allows developer to write an application code, in any programming language, and specify an event to start its execution. Thus, IoT system can get a good benefit of serverless environment. The proposed work introduces a framework to allow Serverless computing to take place on the Fog nodes near the data collectors. The proposed framework is implemented as an extension to a Kubernetes cluster that manages a set of Docker containers at the fog layer. A prototype of the proposed solution was implemented using Node.Js for coding and YAML files to transfer data. The proposed framework was evaluated against traditional cloud Serverless execution. The experimental results proved the significant enhancement of the framework by dcreasing the respond time especially for data intensive IoT applications.