Decentralized Multi-Agent Reinforcement Learning for Task Offloading Under Uncertainty

Yuanchao, Xu; Feriani, Amal; Hossain, Ekram

doi:10.48550/arxiv.2107.08114

Cited by 2 publications

(3 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Attention is used to aggregate the parameter weights resulting in the reduction of the processing time of the task. Yuanchao Xu et al [11] explore decentralized multi-agent reinforcement learning algorithms to solve the problem of task offloading accounting for reward uncertainty. They try different approaches like Multi-Agent Deep-Deterministic Policy-Gradient (MADDPG), Robust MADDPG and Decentralized Partially-Observable Markov Decision Process (Dec-POMDP).…”

Section: Related Workmentioning

confidence: 99%

Actor Critic-based Multi Objective Reinforcement Learning for Multi Access Edge Computing

Khot,

Pai

et al. 2024

IJACSA

View full text Add to dashboard Cite

In recent times, large applications that need near real-time processing are increasingly being used on devices with limited resources. Multi access edge computing is a computing paradigm that provides a solution to this problem by placing servers as close to resource constrained devices as possible. However, the edge device must consider multiple conflicting objectives, viz., energy consumption, latency, task drop rate and quality of experience. Many previous approaches optimize on only one objective or a fixed linear combination of multiple objectives. These approaches don't ensure best performance for applications that run on edge servers, as there is no guarantee that the solution obtained by these approaches lies on the paretofront. In this work, Multi Objective Reinforcement Learning with Actor-Critic model is proposed to optimize the drop rate, latency and energy consumption parameters during offloading decision. The model is compared with MORL-Tabular, MORL-Deep Q Network and MORL-Double Deep Q Network models. The proposed model outperforms all the other models in terms of drop rate and latency.

show abstract

Section: Related Workmentioning

confidence: 99%

Actor Critic-based Multi Objective Reinforcement Learning for Multi Access Edge Computing

Khot,

Pai

et al. 2024

IJACSA

View full text Add to dashboard Cite

show abstract

“…This multi-agent edge computing system models more practical computation offloading scenarios where more than one agent makes decisions together to achieve goals, which may be cooperative or in conflict with each other. Compared to the single-agent computation offloading problem that falls under a category of single-agent RL and can be solved by the popular RL algorithms, like Q-learning and DQN, this multi-agent computation offloading problem falls under the category of multi-agent RL [130].…”

Section: B Multi-agent Rl For Computation Offloadingmentioning

confidence: 99%

“…DNC is a special recurrent neural network and is capable of learning and remembering the past hidden states of inputs. Moreover, the authors in [130] considered the practical challenges of deploying the previously mentioned deep multi-agent RL algorithms and studied applying them to solve task offloading with reward uncertainty.…”

Section: B Multi-agent Rl For Computation Offloadingmentioning

confidence: 99%

Distributed Intelligence in Wireless Networks

Liu

et al. 2023

IEEE Open J. Commun. Soc.

View full text Add to dashboard Cite

The cloud-based solutions are becoming inefficient due to considerably large time delays, high power consumption, and security and privacy concerns caused by billions of connected wireless devices and typically zillions of bytes of data they produce at the network edge. A blend of edge computing and Artificial Intelligence (AI) techniques could optimally shift the resourceful computation servers closer to the network edge, which provides the support for advanced AI applications (e.g., video/audio surveillance and personal recommendation system) by enabling intelligent decision making on computing at the point of data generation as and when it is needed, and distributed Machine Learning (ML) with its potential to avoid the transmission of the large dataset and possible compromise of privacy that may exist in cloud-based centralized learning. Besides, the deployment of AI techniques to redesign end-to-end communication is attracting attention to improve communication performance. Therefore, the interaction of AI and wireless communications generates a new concept, named native AI wireless networks. In this paper, we conduct a comprehensive overview of recent advances in distributed intelligence in wireless networks under the umbrella of native AI wireless networks, with a focus on the design of distributed learning architectures for heterogeneous networks, on AI-enabled edge computing, on the communication-efficient technologies to support distributed learning, and on the AI-empowered end-to-end communications. We highlight the advantages of hybrid distributed learning architectures compared to state-of-the-art distributed learning techniques. We summarize the challenges of existing research contributions in distributed intelligence in wireless networks and identify potential future opportunities.

show abstract

Decentralized Multi-Agent Reinforcement Learning for Task Offloading Under Uncertainty

Cited by 2 publications

References 11 publications

Actor Critic-based Multi Objective Reinforcement Learning for Multi Access Edge Computing

Actor Critic-based Multi Objective Reinforcement Learning for Multi Access Edge Computing

Distributed Intelligence in Wireless Networks

Contact Info

Product

Resources

About