Multiuser context‐aware computation offloading in mobile edge computing based on Bayesian learning automata

• Published in: Transactions on emerging telecommunications technologies Vol. 32; no. 1
• Main Authors: Farahbakhsh, Fariba, Shahidinejad, Ali, Ghobaei‐Arani, Mostafa
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.01.2021
• ISSN: 2161-3915; 2161-3915
• DOI: 10.1002/ett.4127

Today a lot of data sensed from the environment by the Internet of things applications. These data need to process with the lowest delay. Mobile devices (MDs) as ubiquitous tools are end devices in the network. These devices with limited resources cannot process all computations locally. Mobile edge computing (MEC) is a good architecture for processing computations in the network's edge. It solves the cloud challenges such as delay, energy, and cost. If MDs could not process the computations, they will offload tasks to the edge or cloud. Research shows that ignoring context information of application, requests, sensors, resources, and network tools cause to not complete the offloading method. In this article, we consider Bayesian learning automata (BLA) with considering context‐aware offloading in MEC with multiuser. BLA learns all states and actions in the network and helps us to improve the offloading algorithm. The contexts are collected using autonomous management as the monitor‐analysis‐plan‐execution loop in all offloading processes. The simulation results indicate that our method is superior to local computing and offload without considering context‐aware algorithms in some metrics such as energy consumption, execution cost, network usage, delay, and fairness. In this article, we consider Bayesian learning automata (BLA) with considering context‐aware offloading in MEC with multiuser. BLA learns all states and actions in the network and helps us to improve the offloading algorithm. The contexts are collected using autonomous management as the monitor‐analysis‐plan‐execution loop in all offloading processes. The simulation results indicate that the method is superior to local computing and offload without considering context‐aware algorithms in metrics such as energy consumption, execution cost, network usage, delay, and fairness.