Serverless Computing at the Edge for AIoT Applications

Abstract

Serverless computing is a new trend for developing Cloud hosted applications. This paradigm takes advantage of the scalability and flexibility of the management of infrastructure from the Cloud provider to offer advantages in cost and maintenance of Cloud applications. Although it was originally designed for Cloud deployments, some proposals have been made to give its benefits closer where the data is generated and provide service to IoT applications. However, some important challenges remain in order to provide a suitable solution for all cases. This work proposes a computational model for designing a serverless architecture at the edge side of the network. The model is especially focused on Artificial Intelligence (AI) functions of IoT applications, where serverless could achieve its full potential. The results show that the proposal allows to overcome the main drawbacks raised while maintains the cloud processing support.