A DQN-Based Cache Strategy for Mobile Edge Networks

Sun, S.; Zhou, Junhua; Wen, Jiuxing; Wei, Yi; Wang, Xiaojun

doi:10.32604/cmc.2022.020471

Cited by 11 publications

(10 citation statements)

References 35 publications

(38 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The works in [17]- [19] are DQN-based cache allocation approaches, but they simply choose a combination of content to store every step, resulting in huge computational overhead. To alleviate the overhead, the authors fixed the content size uniformly and deployed DQN agents on all nodes with storage.…”

Section: Related Workmentioning

confidence: 99%

Joint Cache Allocation and Replacement for Content-Centric Network-Based Private 5G Networks: Deep Reinforcement Learning Approach

Lim,

Kim,

Yoo

2024

IEEE Access

View full text Add to dashboard Cite

Mobile network traffic volume is increasing significantly every year, which could become a burden on the backbone of the 5G and beyond network in the future. Network caching has been a promising solution to improve quality of service (QoS) by reducing backhaul traffic and network latency. This paper studies network caching for private networks specialized for specific use cases in 5G. We adopt the Content-Centric Network (CCN) as the core network to improve caching efficiency, and suggests a Deep Q-Network (DQN) based method that jointly addresses content allocation strategy and cache replacement policy. Unlike previous studies, the proposed method can handle content and cache storage of heterogeneous sizes by implementing two-bit channel state generator which maps different-sized environment data to a fixed-sized image format. Simulation results shows that the proposed system outperforms existing cache systems by 20% to 50% in terms of cache hit ratio and network latency.

show abstract

Section: Related Workmentioning

confidence: 99%

Joint Cache Allocation and Replacement for Content-Centric Network-Based Private 5G Networks: Deep Reinforcement Learning Approach

Lim,

Kim,

Yoo

2024

IEEE Access

View full text Add to dashboard Cite

show abstract

“…(3) Supply-demand Adaptation Algorithm Based on DRL Powered by a Transfer Mechanism It is necessary to establish a proper external shared parameter pool to share the experiences and conduct transfer learning in order to perform supply-demand adaptation and result optimization. This paper takes the double DQN (DDQN) algorithm [27,28] To verify the model and methods proposed in this paper, two experiments are designed. One aims to show the validity of the divide-and-conquer method for the adaptation domain, and the other aims to verify the effectiveness of the DRL powered by the transfer mechanism in performing the dynamic adaptation task.…”

Section: ) Collaborative Mechanism Of S-d Adaptation Based On Tlmentioning

confidence: 99%

Dynamic Intelligent Supply-Demand Adaptation Model Towards Intelligent Cloud Manufacturing

Sun¹,

Feng²,

Wang³

et al. 2022

Computers, Materials &Amp; Continua

View full text Add to dashboard Cite

As a new mode and means of smart manufacturing, smart cloud manufacturing (SCM) faces great challenges in massive supply and demand, dynamic resource collaboration and intelligent adaptation. To address the problem, this paper proposes an SCM-oriented dynamic supply-demand (S-D) intelligent adaptation model for massive manufacturing services. In this model, a collaborative network model is established based on the properties of both the supply-demand and their relationships; in addition, an algorithm based on deep graph clustering (DGC) and aligned sampling (AS) is used to divide and conquer the large adaptation domain to solve the problem of the slow computational speed caused by the high complexity of spatiotemporal search in the collaborative network model. At the same time, an intelligent supply-demand adaptation method driven by the quality of service (QoS) is established, in which the experiences of adaptation are shared among adaptation subdomains through deep reinforcement learning (DRL) powered by a transfer mechanism to improve the poor adaptation results caused by dynamic uncertainty. The results show that the model and the solution proposed in this paper can perform collaborative and intelligent supply-demand adaptation for the massive and dynamic resources in SCM through autonomous learning and can effectively perform global supply-demand matching and optimal resource allocation.

show abstract

“…Chao Huang et al used a bag-of-words model for modeling and a support vector machine for classification, and after optimizing several global parameters [12]. S. Sun et al started with the network side [13]. Meng Caixia et al followed the filtering method to filter the noise in the fused images and used the calibrated approximate rectangular response technique to delineate the rioters' areas to achieve accurate detection of rioters [14].…”

Section: Related Workmentioning

confidence: 99%

A Novel Transfer Learning-Based Algorithm for Detecting Violence Images

2022

KSII TIIS

View full text Add to dashboard Cite

Violence in the Internet era poses a new challenge to the current counter-riot work, and according to research and analysis, most of the violent incidents occurring are related to the dissemination of violence images. The use of the popular deep learning neural network to automatically analyze the massive amount of images on the Internet has become one of the important tools in the current counter-violence work. This paper focuses on the use of transfer learning techniques and the introduction of an attention mechanism to the residual network (ResNet) model for the classification and identification of violence images. Firstly, the feature elements of the violence images are identified and a targeted dataset is constructed; secondly, due to the small number of positive samples of violence images, pre-training and attention mechanisms are introduced to suggest improvements to the traditional residual network; finally, the improved model is trained and tested on the constructed dedicated dataset. The research results show that the improved network model can quickly and accurately identify violence images with an average accuracy rate of 92.20%, thus effectively reducing the cost of manual identification and providing decision support for combating rebel organization activities.

show abstract

A DQN-Based Cache Strategy for Mobile Edge Networks

Cited by 11 publications

References 35 publications

Joint Cache Allocation and Replacement for Content-Centric Network-Based Private 5G Networks: Deep Reinforcement Learning Approach

Joint Cache Allocation and Replacement for Content-Centric Network-Based Private 5G Networks: Deep Reinforcement Learning Approach

Dynamic Intelligent Supply-Demand Adaptation Model Towards Intelligent Cloud Manufacturing

A Novel Transfer Learning-Based Algorithm for Detecting Violence Images

Contact Info

Product

Resources

About