Machine learning for cloud, fog, edge and serverless computing environments: comparisons, performance evaluation benchmark and future directions

The compute-intensive and latency-sensitive Internet of Things (IoT) applications need to use the services from various computing paradigms, but they are facing many challenges such as large values of latency, energy and network bandwidth. To analyse and understand these challenges, we designed a performance evaluation benchmark that integrates cloud, fog, edge and serverless computing to conduct a comparative study for IoT-based healthcare applications. It gives the platform for the developers to design IoT applications based on user guidelines to run various applications concurrently on different paradigms. Furthermore, we used recent machine learning techniques for the optimisation of resources, energy, cost and overheads to identify the best technique based on important Quality of Service (QoS) parameters. Experimental results show that serverless computing performs better than non-serverless in terms of energy, latency, bandwidth, response time and scalability by 3.8%, 3.2%, 4.3%, 1.5% and 2.7%, respectively. Finally, various promising future directions are highlighted