Application of Machine Learning Algorithm in Cloud-to-edge Computing: Analysis and Limitations

Machine learning (ML) approaches have been shown to be useful in a variety of difficult issues and domains such as managing resources, cloud services, and edge computing when used appropriately. Many cloud services concepts exist, including edge computing, fog computing, and mist computing, Internet of Things (IoT), Software-Defined Networking (SDN), cyber twin, and industry 4.0, have emerged in response to client application needs. These paradigms work together to provide customer-centric services delivered through the cloud server/data centers backend. However, a full review focused on cloud-to-edge computational resources management, technical and analytical features of these concepts, and the role of ML approaches in new cloud-to-edge computing concepts is still lacking, and this topic needs to be researched. As a result, this paper surveys the rising cloud-to-edge computing paradigms integration while taking into account the most dominant problem-solving technology by ML. This study provides a complete literature assessment of new cloud-to-edge computing paradigms, as well as their integration with ML. To carry out this research, articles from the last ten years (2013–2022) are extensively explored and analyzed in order to comprehend the application of various ML approaches in resource management of the cloud-to-edge ecosystem, and the comparison analysis on numerous aspects, including recent developments. Lastly, based on the observed numerous ML resource management issues and deficiencies in present methodologies for addressing these challenges, the paper proposes probable future research topics.