Artificial Intelligence Techniques for Dynamic Offloading in Cloud Continuum Environment: A Review

In this work we are going to present the most recent progress on the topic of dynamic offloading guided by artificial intelligence (AI) techniques. The growing demands in terms of performance on edge devices often require very high computational capabilities. In spite of the progress made in hardware power, some applications, such as collaborative augmented reality (AR) and virtual reality (VR) or particular needs in medical or military sectors, require rapid response times which put a strain on the performance of the devices. In these contexts, the use of dynamic offloading in a fog and edge computing environment emerge an effective solution to increase the performance of the architectures. In the first part of this review we aim to analyze the enabling technologies, objectives, strategies and opportunities of dynamic offloading. After, based on this investigation and with the analysis of different algorithms developed, we would like to show how artificial intelligence can play a crucial role in this context, in particular with the use of machine learning (ML), deep learning (DL) or reinforcement learning (RL) models also with the use of hybrid solutions.