Decentralized Task Offloading in Edge Computing: A Multi-User Multi-Armed Bandit Approach

Xiong, Wei; Ye, Jiancheng; Lui, John C. S.

doi:10.1109/infocom48880.2022.9796961

Cited by 54 publications

(6 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Then, the exploitation phase is the phase where the agent continuously chooses the action having the highest reward for future plays. Many studies resort to the multi-armed bandits (MAB), a RL framework applied in a single-state environment [9], to solve the problem of channel allocation in ad-hoc networks and treat the exploration and exploitation phases separately to learn the channels [10] and to manage the strategy for edge users [11].…”

Section: Prerequisites and System Componentsmentioning

confidence: 99%

Uncoordinated Spectrum Access for Massive Connectivity in IoT Applications

Doumit,

Youssef,

Nour

et al. 2024

IEEE Internet Things M.

View full text Add to dashboard Cite

Boosted by the Internet of Things (IoT), future cellular communication networks will be able to support new services with high social impact. Corresponding applications are expected to span almost every domain, especially the most critical ones, such as the medical field or smart factories. However, their adoption is conditioned by strict compliance with stringent quality of service (QoS) requirements for the underlying network. In particular, the achieved throughput and latency are of utmost importance for efficient monitoring with multiple sensors in a variety of application areas. In this paper, we propose an intelligent self-configurable system structure, achieved through resource allocation techniques using machine learning, and capable of flexibly fulfilling such requirements. In addition, special care was taken to minimize the transmit power of the IoT devices while efficiently supporting different application-dependent traffic types, an essential prerequisite for any wide adoption of such systems with massive network access.

show abstract

Section: Prerequisites and System Componentsmentioning

confidence: 99%

Uncoordinated Spectrum Access for Massive Connectivity in IoT Applications

Doumit,

Youssef,

Nour

et al. 2024

IEEE Internet Things M.

View full text Add to dashboard Cite

show abstract

“…Mobile Edge Computing: Existing excellent works have conducted various research questions in MEC, including re-source allocation (e.g., (Wang et al 2022b)), service placement (e.g., (Taka, He, and Oki 2022)), and proactive caching (e.g., (Liu et al 2022a)). Task offloading (Wang, Ye, and Lui 2022;Ma et al 2022;Chen and Xie 2022), as another main research question in MEC, attracting considerable attention.…”

Section: Literature Reviewmentioning

confidence: 99%

Fractional Deep Reinforcement Learning for Age-Minimal Mobile Edge Computing

Jin,

Tang,

Zhang

et al. 2024

AAAI

View full text Add to dashboard Cite

Mobile edge computing (MEC) is a promising paradigm for real-time applications with intensive computational needs (e.g., autonomous driving), as it can reduce the processing delay. In this work, we focus on the timeliness of computational-intensive updates, measured by Age-of-Information (AoI), and study how to jointly optimize the task updating and offloading policies for AoI with fractional form. Specifically, we consider edge load dynamics and formulate a task scheduling problem to minimize the expected time-average AoI. The uncertain edge load dynamics, the nature of the fractional objective, and hybrid continuous-discrete action space (due to the joint optimization) make this problem challenging and existing approaches not directly applicable. To this end, we propose a fractional reinforcement learning (RL) framework and prove its convergence. We further design a model-free fractional deep RL (DRL) algorithm, where each device makes scheduling decisions with the hybrid action space without knowing the system dynamics and decisions of other devices. Experimental results show that our proposed algorithms reduce the average AoI by up to 57.6% compared with several non-fractional benchmarks.

show abstract

“…The simulation results further affirmed that the newly proposed algorithm can significantly conserve overhead compared to other existing computation offloading algorithms. Other inquiries documented in [18][19][20] addressed similar multi-user and multi-server environments. These studies introduced heuristic decision-making algorithms for offloading, which strategically assigned different tasks to optimal channels and servers with the objective of reducing average completion times.…”

Section: Related Workmentioning

confidence: 99%

Probabilistic Task Offloading with Uncertain Processing Times in Device-to-Device Edge Networks

Shu,

Luo,

Liu

2024

Electronics

View full text Add to dashboard Cite

D2D edge computing is a promising solution to address the conflict between limited network capacity and increasing application demands, where mobile devices can offload their tasks to other peer devices/servers for better performance. Task offloading is critical to the performance of D2D edge computing. Most existing works on task offloading assume the task processing time is known or can be accurately estimated. However, the processing time is often uncertain until it is finished. Moreover, the same task can have largely different execution times under different scenarios, which leads to inaccurate offloading decisions and degraded performance. To address this problem, we propose a game-based probabilistic task offloading scheme with an uncertain processing time in D2D edge networks. First, we characterize the uncertainty of the task processing time using a probabilistic model. Second, we incorporate the proposed probabilistic model into an offloading decision game. We also analyze the structural properties of the game and prove that it can reach a Nash equilibrium. We evaluate the proposed work using real-world applications and datasets. The experimental results show that the proposed probabilistic model can accurately characterize the uncertainty of completion time, and the offloading algorithm can effectively improve the overall task completion rate in D2D networks.

show abstract

Decentralized Task Offloading in Edge Computing: A Multi-User Multi-Armed Bandit Approach

Cited by 54 publications

References 37 publications

Uncoordinated Spectrum Access for Massive Connectivity in IoT Applications

Uncoordinated Spectrum Access for Massive Connectivity in IoT Applications

Fractional Deep Reinforcement Learning for Age-Minimal Mobile Edge Computing

Probabilistic Task Offloading with Uncertain Processing Times in Device-to-Device Edge Networks

Contact Info

Product

Resources

About