Resource allocation problem and artificial intelligence: the state‑of‑the‑art review (2009–2023) and open research challenges

Abstract

With the increasing growth of information through smart devices, enhancing the quality of human life necessitates the adoption of various computational paradigms including, cloud, fog, and edge in the Internet of Things (IoT) network. Among these paradigms, cloud computing, as an emerging technology, extends cloud layer services to the network edge. This enables resource allocation operations to occur closer to the end user, thereby reducing resource processing time and network traffic overhead. Consequently, the resource allocation problem for cloud service providers, in terms of presenting a suitable platform using computational paradigms, is considered a challenge. Also, Energy Efficiency, Heterogeneity, and Scalability are problems of the cloud computing environment. To solve these problems, resource allocation approaches are divided into two methods: auction-based methods (aimed at increasing profits for service providers while ensuring user satisfaction and usability) and optimization-based methods (focused on energy, cost, network exploitation, runtime, and time delay reduction). Hence, this paper presents a comprehensive literature study (CLS) on artificial intelligence methods such as machine and deep learning for resource allocation optimization in computing environments, such as cloud computing, fog computing, and edge computing. Since deep learning methods are widely used in resource allocation problems, this paper also explores resource allocation approaches based on deep learning techniques, such as deep reinforcement learning, Q-learning, reinforcement learning, and online learning, as well as classical learning methods like Bayesian learning. As a main important achievement, the use of deep reinforcement learning-based methods has increased in the fog paradigm in the past few years.