Hybrid Reinforcement Learning based code offloading in MEC

Pati, Abhinash (2020) Hybrid Reinforcement Learning based code offloading in MEC. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (887kB) | Preview

Abstract

In recent times, the Mobile Edge Computing paradigm has become a popular paradigm in latency reduction for resource-constrained devices by utilizing the concept of code offloading to another execution platform. With this, the IoT devices have got a boot in data processing capabilities. This field is being widely researched; however, many issues persist. Although many method have been developed to get real-time results from such paradigms, it still not perfect. A variety of unpredictable constraints such as device roaming and unreliable network conditions hinders the optimal operation of the code offloading, resulting in delays. This paper proposes an online algorithm using deep reinforcement learning and sampling and classification approach to find the most suitable execution node in an execution platform. The primary purpose of the algorithm is to obtain the best fit node where if the task is offloaded, the total cost of execution (TCE) will be minimum. The algorithm considers the Femto cloud-based mobile cluster as a potential platform for task offloading. It implements a Deep Q-Network which learns from the error generated during the decision-making process. The results, which are based on publicly available datasets used to simulate the proposed system, show that the proposed algorithm performs better than the baseline algorithms without any learning component.

Item Type:	Thesis (Masters)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Dan English
Date Deposited:	28 Jan 2021 17:11
Last Modified:	28 Jan 2021 17:11
URI:	https://norma.ncirl.ie/id/eprint/4546

Actions (login required)

View Item