Transfer Learning Promotes 6G Wireless Communications: Recent Advances and Future Challenges

Wang, Meiyu; Lin, Yun; Tian, Qiao; Si, Guangzhen

doi:10.1109/tr.2021.3062045

Cited by 175 publications

(58 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One article [78] described, as mentioned above, requirements were contradictory, such as high efficiency and high density. This combination eventually led to reduced energy efficiency.…”

Section: Review and Survey Articlesmentioning

confidence: 99%

From 5G to 6G Technology: Meets Energy, Internet-of-Things and Machine Learning: A Survey

Mahdi

Ahmad

Qassim³

et al. 2021

Applied Sciences

View full text Add to dashboard Cite

Due to the rapid development of the fifth-generation (5G) applications, and increased demand for even faster communication networks, we expected to witness the birth of a new 6G technology within the next ten years. Many references suggested that the 6G wireless network standard may arrive around 2030. Therefore, this paper presents a critical analysis of 5G wireless networks’, significant technological limitations and reviews the anticipated challenges of the 6G communication networks. In this work, we have considered the applications of three of the highly demanding domains, namely: energy, Internet-of-Things (IoT) and machine learning. To this end, we present our vision on how the 6G communication networks should look like to support the applications of these domains. This work presents a thorough review of 370 papers on the application of energy, IoT and machine learning in 5G and 6G from three major libraries: Web of Science, ACM Digital Library, and IEEE Explore. The main contribution of this work is to provide a more comprehensive perspective, challenges, requirements, and context for potential work in the 6G communication standard.

show abstract

“…One article [78] described, as mentioned above, requirements were contradictory, such as high efficiency and high density. This combination eventually led to reduced energy efficiency.…”

Section: Review and Survey Articlesmentioning

confidence: 99%

From 5G to 6G Technology: Meets Energy, Internet-of-Things and Machine Learning: A Survey

Mahdi

Ahmad

Qassim³

et al. 2021

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…As shown in the table, we have evaluated various state-of-theart DRL algorithms implemented in the Tensorforce Python package 2 . We mainly tested three classes of DRL algorithms, namely Deep Q-Network (DQN), Actor Critic (AC), and Proximal Policy Optimization (PPO).…”

Section: A Simulation Environment Settingsmentioning

confidence: 99%

“…Such challenges are rarely tackled in the wireless networks literature or research. On a related note, transfer learning (TL) has recently achieved some noticeable results in the wireless communications domain [2]. TL is a paradigm that generally focuses on reusing knowledge gained while solving a learning task.…”

Section: Introductionmentioning

confidence: 99%

Transfer Learning-Based Accelerated Deep Reinforcement Learning for 5G RAN Slicing

Nagib

Abou-zeid

Hassanein

2021

2021 IEEE 46th Conference on Local Computer Networks (LCN)

View full text Add to dashboard Cite

Deep Reinforcement Learning (DRL) algorithms have been recently proposed to solve dynamic Radio Resource Management (RRM) problems in 5G networks. However, the slow convergence experienced by traditional DRL agents puts many doubts on their practical adoption in cellular networks. In this paper, we first discuss the need to have accelerated DRL algorithms. We then analyze the exploration behavior of various state-of-the-art DRL algorithms for slice resource allocation, and compare it with the traditional 5G Radio Access Network (RAN) slicing baselines. Finally, we propose a transfer learningaccelerated DRL-based solution for slice resource allocation. In particular, we tackle the challenge of slow convergence by transferring the policy learned by a DRL agent at an expert base station (BS) to newly deployed agents at target learner BSs. Our approach shows a remarkable reduction in convergence time and a significant performance improvement compared with its nonaccelerated counterparts when tested against multiple traffic load variations.

show abstract

“…Nowadays, automobile traffic has become an indispensable part of our modern industry, and vehicle-related technologies and industries are becoming increasingly mature [1][2][3][4]. At the same time, the development of technologies related to the Internet of Vehicles is also promoting the evolution of the traditional automobile industry to the new intelligent vehicles, and the market demand and technological exploration are mutually reinforce each other, including industrial transportation, advanced traffic information system, and travel technology, which can provide navigation and positioning, road status, parking instructions, safe driving assistance, and other all-round services [5][6][7][8].…”

Section: Introductionmentioning

confidence: 99%

“…Borrowing the cloud platform, we can estimate their DOA and locate these vehicles according to geometrical relationship of the signals and the infrastructure. It not only improves the accuracy but also enhances the resolution for the target, which is very suitable for multiple vehicles (2) In this paper, signal information of different frequencies is extracted as input feature, while DOA is taken as the output for training; then, DOA of vehicle is estimated by RFR; the prediction model can be adjusted through parameter optimization. This process is more intuitive and without complex parameter settings and has good robustness, scalability, and flexibility.…”

Section: Introductionmentioning

confidence: 99%

Accessorial Locating for Internet of Vehicles Based on DOA Estimation in Industrial Transportation

Zhen

Guo

2021

Wireless Communications and Mobile Computing

View full text Add to dashboard Cite

With the improvement of living standard and the development of science and technology, Internet of Vehicle (IOV) will play an important part in industrial transportation as a main research field of Internet of Things. As a result, it is very necessary to grasp the location of vehicle. However, the traditional single global position system is easily affected by the external environment, so an accessorial locating approach based on wideband direction of arrival (DOA) estimation in intelligent transportation is proposed. First, model the array received signal on the road infrastructure. Then, by means of random forest regression (RFR) in the supervised learning, upper triangle elements of the covariance matrix of each frequency and the actual DOA are, respectively, extracted as the input features and output parameters; thus, the corresponding prediction coefficients are solved by training. After that, the trained RFR model can be used to calculate the final direction using test samples. Finally, these vehicles can be located according to the geometrical relation between the vehicle and the infrastructure. The proposed algorithm is not only suitable for uncorrelated signals but also for uncorrelated and correlated mixed signals without wideband focusing. The simulations show that compared with some sparse recovery algorithm, the prediction accuracy and resolution are effectively improved.

show abstract

Transfer Learning Promotes 6G Wireless Communications: Recent Advances and Future Challenges

Cited by 175 publications

References 75 publications

From 5G to 6G Technology: Meets Energy, Internet-of-Things and Machine Learning: A Survey

From 5G to 6G Technology: Meets Energy, Internet-of-Things and Machine Learning: A Survey

Transfer Learning-Based Accelerated Deep Reinforcement Learning for 5G RAN Slicing

Accessorial Locating for Internet of Vehicles Based on DOA Estimation in Industrial Transportation

Contact Info

Product

Resources

About