Game-theoretic methods for edge/fog/cloud computation offloading: a systematic review

Hassan Pirhosseini, Edris Zarei, Mohammad Hossein Rezvani, Azadeh Pourkabirian, Achyut Shankar, Nebojsa Bacanin, Carsten Maple, Wattana Viriyasitavat

Abstract: Due to the distance of users from remote cloud data centers, cloud computing (CC) may not be a good solution for latency-sensitive applications. Fog computing (FC) and Mobile Edge Computing (MEC) are two complementary paradigms for CC that can reduce energy consumption and delay. In FC, if the cloud resources are not enough to process the task, it is offloaded to the remote cloud. On the other hand, the main advantage of MEC is to reduce latency and availability of data and services for end users. If the edge resources are insufficient to process the work, it may be offloaded to remote cloud servers or fog. One of the suitable tools for modeling edge/fog/cloud discharge is Game Theory (GT). Game theory tries to predict the behaviors and decision results of agents who have the right to choose in interaction with each other by using scenario design and analysis. Here, the goal is to achieve a stable allocation of network resources to meet user requirements. Game-theoretic modeling can lead to agents located at the edge of the network acting more reliably. In this paper, we present an extensive systematic review of GT-based task offloading in different computational paradigms such as MEC, FC, and MCC. In addition to classical game theory, we pay special attention to evolutionary game theory. Generally, these methods are more scalable than classical methods. Instead of considering the user as a player, they consider network characteristics such as the number of CPU cycles. This article applies game theory from different perspectives such as the type of offloading, performance criteria, algorithm type, system components, evaluation tools, and computing environment.

External IDs:dblp:journals/computing/PirhosseiniZRPSBMV25