An Edge Federated MARL Approach for Timeliness Maintenance in MEC Collaboration
Authors:
- 1 June 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Abstract
Mobile edge computing (MEC) has been widely studied to provide new schemes for communication-computing systems such as industrial Internet of Things (IoTs), vehicular networks, smart city applications, etc. In this work, we mainly investigate on the timeliness maintenance of the MEC systems where the freshness of the data and computation tasks plays a significant role. We firstly formulate the average age of information (AoI) minimization problem of the UAV-assisted MEC systems. To maintain the system timeliness, we propose a novel multi-agent reinforcement learning (MARL) approach, called edge federated multi-agent actor-critic (MAAC), for joint trajectory planning, data scheduling and resource management in the investigated MEC systems. Through the proposed online learning method, edge devices and center controller learn the interactive policies through local observations and carry out the model-wise communication. We build up a simulation platform for time sensitive MEC systems as a gym environment module and implement the proposed algorithm. Furthermore, the comparisons with a popular MARL solution, MADDPG, show that the proposed approach achieves better performance in terms of data freshness and system stability.Keywords
This publication has 16 references indexed in Scilit:
- Machine Learning Based Prediction and Classification of Computational Jobs in Cloud Computing Centers2019
- Joint Communication, Computation, Caching, and Control in Big Data Multi-Access Edge ComputingIEEE Transactions on Mobile Computing, 2019
- Caching and Computing at the Edge for Mobile Augmented Reality and Virtual Reality (AR/VR) in 5G2018
- Latency and Reliability-Aware Task Offloading and Resource Allocation for Mobile Edge Computing2017
- A Survey on Mobile Edge Computing: The Communication PerspectiveIEEE Communications Surveys & Tutorials, 2017
- Edge Computing: Vision and ChallengesIEEE Internet of Things Journal, 2016
- Human-level control through deep reinforcement learningNature, 2015
- Optimal LAP Altitude for Maximum CoverageIEEE Wireless Communications Letters, 2014
- Internet of Things (IoT): A vision, architectural elements, and future directionsFuture Generation Computer Systems, 2013
- A Comprehensive Survey of Multiagent Reinforcement LearningIEEE Transactions on Systems, Man and Cybernetics, Part C (Applications and Reviews), 2008