The ever-growing big data has empowered ubiquitous artificial intelligence (AI) for a wide range of novel technologies that keep reshaping our world to evolve from Internet of Everything (IoE) to Internet of Intelligence (IoI). Current deployment 5G networks, however, relies on centralized management and model-based analytics over fix-configured infrastructures, and thus exposes limitations for achieving seamless connections among humans and ubiquitous things in terms of unlimited connectivity, space-air-ground coverage, extremely low latency, strong robustness, and self-adaptation capability. These features are missing from the status quo. However, in expected beyond 5G (B5G) and 6G networks, the launch of the next-generation network system with revolutionary technologies (such as intelligent surfaces, cell-free architecture, massive MIMO, space-air-ground integrated networks, and cyber-twin) will be able to transition the role of access networks from the carrier of data to the carrier of intelligence, so that the advance of machine learning-enabled paradigms can be accelerated over the synergy of smart networking and connected intelligence.
The reciprocal relationship between next-generation access network and future AI development has encouraged many new research areas, including AI-enabled network optimization, distributed system, edge computing, federated learning at scale, and peer-to-peer learning. Status quo literatures often have limitations due to the legacy-configured network constraints, or impractical assumptions for idealized data communication, and thus lack feasibility in practical systems. In recent years, there has been great progress in novel AI analytics and intelligent paradigms that greatly improved the learning efficiency and accuracy. Simply applying these learning paradigms to wireless network scenarios can often ignore the actual challenges in wireless communication and miss the unique characteristics and technical advantages of new networking solutions, and thus restrict the further application of the latest progress in machine-learning technologies.
To promote the learning intelligence for mobile users distributed in broad geographical scope, and to emphasize its significance in a future-generation network, this special section calls for novel and promising machine-learning paradigms that take advantage of new features of B5G/6G networks that can be widely applied in many scenarios (such as vehicular networks, large-scale Internet of Things, and space-air-ground integrated networks). The special section aims to invite researchers from both academia and industry to present their research findings and engineering practices to envision a future that ubiquitous AI can be provisioned via the interdisciplinary research between computing and communication. The special section seeks original and prominent research works on state-of-the-art learning approaches, methodologies, and key technologies, as well as practical systems regarding ubiquitous learning over a new generation of wireless access networks.
Topics of interest include, but are not limited to:
- Emerging learning architecture and framework over B5G/6G networks
- Federated learning over large-scale Internet of Things
- Collaborative learning over vehicular networks
- Knowledge-driven paradigms for future networking
- Cyber-twin-driven 6G for ubiquitous machine learning
- Ubiquitous learning in space-air-ground integrated networks
- Machine learning in cloud-edge networks
- Machine-learning techniques in sensing-communication-computing integration
- Meta learning for knowledge transfer over B5G/6G networks
- Transfer learning in heterogeneous networks
- AI-enabled intelligent networking in B5G/6G networks
- Adversary learning for security and privacy preserving in B5G/6G networks
- Reinforcement learning for intelligent strategy in 6G communication
Important Dates
Submissions due: May 15, 2021
Notification to authors: June 15, 2021
Revisions due: July 15, 2021
Final notification: August 15, 2021
Publication: 2021
Guest Editors
Wenchao Xu, The Hong Kong Polytechnic University, Hong Kong, wenchao.xu@polyu.edu.hk
Cunqing Hua, Shanghai Jiao Tong University, China, cqhua@sjtu.edu.cn
Haibo Zhou, Nanjing University, China, haibozhou@nju.edu.cn
Nan Cheng, Xidian University, China, dr.nan.cheng@ieee.org
Mehrdad Dianati, University of Warwick, UK, M.Dianati@warwick.ac.uk