High-Reliability Multi-Agent Q-Learning-Based Scheduling for D2D Microgrid Communications

Shimotakahara, Kevin; Elsayed, Medhat; Hinzer, Karin; Erol‐Kantarci, Melike

doi:10.1109/access.2019.2920662

Cited by 11 publications

(8 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…From the perspective of grid tasks, the optimization model introduces the time constraints in the traditional grid task scheduling algorithm into the Ad Hoc grid energy optimization. By reducing the task completion time MET, to achieve The purpose of energy optimization [5][6]. Therefore, the scheduling model is based on the traditional energy optimization model by adding the factor of time cost, and comprehensively considering the time factor and energy factor to realize the energy optimization of scheduling [6][7].…”

Section: Energy Optimization Modelmentioning

confidence: 99%

Energy Optimization Based on Grid Resource Scheduling Algorithm

2021

View full text Add to dashboard Cite

Grid technology is an emerging technology developed in the mode of computer network computing. This technology has many characteristics such as distribution, sharing, and polymorphism. In the grid environment, due to the high-performance computing of the grid, the task scheduling process becomes efficient, but it also has the problem of complex grid resource management and scheduling strategies, resulting in huge energy consumption. In order to solve the problem of energy consumption, an energy optimization model based on time constraints and energy constraints is proposed in this paper, grid resource scheduling is carried out through heuristic scheduling algorithm, and energy optimization simulation experiments are carried out under the condition of changing the number of resources and tasks. The results show that, The resource execution time corresponding to a single grid task is short, and the energy consumption value is also small. In the simulation experiment of multiple grid tasks, as the number of grid tasks increases, the task execution time increases, and the adjustment factor is 0.5 , that is, when the ratio of the time consumption factor and the energy consumption factor in the resource scheduling optimization cost function is the same, the fluctuation of the energy consumption rate is relatively stable.

show abstract

Section: Energy Optimization Modelmentioning

confidence: 99%

Energy Optimization Based on Grid Resource Scheduling Algorithm

2021

View full text Add to dashboard Cite

show abstract

“…5, each hidden layer is composed of a fully connected (FC) sublayer and a rectified linear unit (ReLU) sublayer in series. The ReLU sublayer improves calculation speed and prediction accuracy by retaining the positive values and eliminating the (15)…”

Section: Multipath Dnn Trainingmentioning

confidence: 99%

“…Most of the current research on RL methods in wireless resource allocation is based on the action-value function to approximate the optimal action selection [14][15][16]. In [14], the authors present a software-defined satellite-terrestrial network framework, which jointly considered the networking, caching, and computing resources in satelliteterrestrial networks.…”

mentioning

confidence: 99%

“…A deep Q-learning was used to approximate the optimal expected utility of each resource. The authors in [15] proposed a multiagent Q-learning resource management algorithm, which reduced the packet loss rate and power oscillation in the device-to-device (D2D) network. In [16], a Q-learning cooperative power allocation algorithm was proposed to increase the capacity of two-tier dense heterogeneous networks (HetNets).…”

mentioning

confidence: 99%

See 1 more Smart Citation

Reinforcement learning-based hybrid spectrum resource allocation scheme for the high load of URLLC services

Huang

Xie

Cheriet

2020

J Wireless Com Network

View full text Add to dashboard Cite

Ultra-reliable and low-latency communication (URLLC) in mobile networks is still one of the core solutions that require thorough research in 5G and beyond. With the vigorous development of various emerging URLLC technologies, resource shortages will soon occur even in mmWave cells with rich spectrum resources. As a result of the large radio resource space of mmWave, traditional real-time resource scheduling decisions can cause serious delays. Consequently, we investigate a delay minimization problem with the spectrum and power constraints in the mmWave hybrid access network. To reduce the delay caused by high load and radio resource shortage, a hybrid spectrum and power resource allocation scheme based on reinforcement learning (RL) is proposed. We compress the state space and the action space by temporarily dumping and decomposing the action. The multipath deep neural network and policy gradient method are used, respectively, as the approximater and update method of the parameterized policy. The experimental results reveal that the RL-based hybrid spectrum and the power resource allocation scheme eventually converged after a limited number of iterative learnings. Compared with other schemes, the RL-based scheme can effectively guarantee the URLLC delay constraint when the load does not exceed 130%.

show abstract

“…It has been deeply studied because it has the advantages of simple implementation and strong convergence. For example, in [16] the authors proposed the multi-agent Q-learning algorithm to reduce packet drop rates in D2D communication. Huang et al [17] proposed a distributed Q-learning algorithm in heterogeneous networks to minimise the total transmission power.…”

Section: Introductionmentioning

confidence: 99%