iRAF: A Deep Reinforcement Learning Approach for Collaborative Mobile Edge Computing IoT Networks

Chen, Jienan; Chen, Siyu; Wang, Qi; Cao, Bin; Feng, Gang; Hu, Jianhao

doi:10.1109/jiot.2019.2913162

Cited by 190 publications

(61 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reference [36] used DDQN on virtual edge computing and obtained good performance in offloading computing. The literature [37] applied the Monte Carlo tree search algorithm to the resource allocation of MEC, and the performance of the scheme was significantly better than that of DQN. Reference [38] combined the computing offload of edge computing with block chain, comprehensively considered the time delay energy consumption and the cost of the block chain, and achieved a good effect with the DRL-based algorithm.…”

Section: Related Workmentioning

confidence: 99%

Distributed Edge Computing Offloading Algorithm Based on Deep Reinforcement Learning

Wang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

As a mode of processing task request, edge computing paradigm can reduce task delay and effectively alleviate network congestion caused by the proliferation of Internet of things(IoT) devices compared with cloud computing. However, in the actual construction of the network, there are various edge autonomous subnets in the adjacent areas, which leads to the possibility of unbalance of server load among autonomous subnets during the peak period of task request. In this paper, a deep reinforcement learning algorithm is proposed to solve the complex computation offloading problem for the heterogeneous Edge Computing Server(ECS) collaborative computing. The problem is solved based on the real-time state of the network and the attributes of the task, which adopts Actor Critic and Policy Gradient's Deep Deterministic Policy Gradient(DDPG) to make optimized decisions of computation offloading. Considering multi-task, the heterogeneity of edge subnet and mobility of edge devices, the proposed algorithm can learn the network environment and generate the computation offloading decision to minimize the task delay.The simulation results show that the proposed DDPG-based algorithm is competitive compared with the Deep Q Network(DQN) algorithm and Asynchronous Advantage Actor-Critic(A3C) algorithm. Moreover, the optimal solutions are leveraged to analyze the influence of edge network parameters on task delay. INDEX TERMS Edge computing, computation offload, collaborative computing, reinforcement learning, DDPG.

show abstract

Section: Related Workmentioning

confidence: 99%

Distributed Edge Computing Offloading Algorithm Based on Deep Reinforcement Learning

Wang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Generally, to achieve better results, other heuristic algorithms and pre-trained neural networks are needed. In the work of Chen et al, the DQN-based method should learn a policy by experience transitions with a well-trained Deep Neural Network(DNN) [18]. In order to optimize computational offloading under our settings with high performances, we apply Dueling DQN, an improved DQN method for resource resign and allocation in MEC system.…”

Section: Related Workmentioning

confidence: 99%

Semi-Online Computational Offloading by Dueling Deep-Q Network for User Behavior Prediction

et al. 2020

View full text Add to dashboard Cite

Task offloading could optimize computational resource utilization in edge computing environments. However, how to assign and offload tasks for different behavior users is an essential problem since the systems dynamic, intelligent application diversity, and user personality. With user behavior prediction, this paper proposes soCoM, a semi-online Computational Offloading Model. We explore the user behaviors in sophisticated action space by reinforcement learning for catching unknown environment information. With Dueling Deep-Q Network, both the prediction accuracy of users' behaviors and the server load balance are well-considered, while increasing the computational efficiency and decreasing the resource costing. We propose a dynamic simulation environment of edge computing to demonstrate that user behavior is the critical factor for impacting system utilization. As the action space increasing, Dueling DQN performs better than state-of-art DQN and other improved strategies, and also load balance in multiple different server scenario.

show abstract

“…To reduce the collisions between the system and client effectively, the access method is designed using rule-based algorithms and RL. RL is also utilized as a means of selecting the appropriate channel [16][17][18][19]. RL is applied to the problem of choosing a route to escape to a destination by avoiding obstacles.…”

Section: Learning From Demonstrationmentioning

confidence: 99%

Adaptive Human–Machine Evaluation Framework Using Stochastic Gradient Descent-Based Reinforcement Learning for Dynamic Competing Network

Kim

Lee

2020

Applied Sciences

View full text Add to dashboard Cite

Complex problems require considerable work, extensive computation, and the development of effective solution methods. Recently, physical hardware- and software-based technologies have been utilized to support problem solving with computers. However, problem solving often involves human expertise and guidance. In these cases, accurate human evaluations and diagnoses must be communicated to the system, which should be done using a series of real numbers. In previous studies, only binary numbers have been used for this purpose. Hence, to achieve this objective, this paper proposes a new method of learning complex network topologies that coexist and compete in the same environment and interfere with the learning objectives of the others. Considering the special problem of reinforcement learning in an environment in which multiple network topologies coexist, we propose a policy that properly computes and updates the rewards derived from quantitative human evaluation and computes together with the rewards of the system. The rewards derived from the quantitative human evaluation are designed to be updated quickly and easily in an adaptive manner. Our new framework was applied to a basketball game for validation and demonstrated greater effectiveness than the existing methods.

show abstract

iRAF: A Deep Reinforcement Learning Approach for Collaborative Mobile Edge Computing IoT Networks

Cited by 190 publications

References 43 publications

Distributed Edge Computing Offloading Algorithm Based on Deep Reinforcement Learning

Distributed Edge Computing Offloading Algorithm Based on Deep Reinforcement Learning

Semi-Online Computational Offloading by Dueling Deep-Q Network for User Behavior Prediction

Adaptive Human–Machine Evaluation Framework Using Stochastic Gradient Descent-Based Reinforcement Learning for Dynamic Competing Network

Contact Info

Product

Resources

About