Call for Papers

With the development of hardware and artificial intelligence (AI) technologies, more and more wireless devices are equipped with various levels of computing hardware, enabling distributed AI via node collaboration. Compared with centralized AI which are often realized on cloud servers, distributed AI can better exploit and coordinate the computation, storage, and communication resources on devices, to realize the crowdsourcing of data as well as the machine learning (ML) tasks. As a result, infrastructure basis becomes less crucial and privacy is better protected which is crucial for both model training and inference in ML applications. In addition, with the help from nodes in the proximity rather than cloud servers, distributed AI can enable just-in-time AI support, especially for inference. Thanks to these advantages, distributed AI has the potential to trigger paradigm change for wireless technologies from model-based to data-driven. On the other hand, distributed AI can boost the development of many emerging applications, such as autonomous driving, augment and virtual reality, and industrial Internet.

Distributed AI inevitably consumes notable energy from wireless communications for frequent information exchange among participating nodes, as well as computing energy from running machine learning algorithms. However, wireless devices are often energy constrained, powered by battery or renewable energy. This conflict has grown into one of the most important issues to enable sustainable distributed AI, and thus strongly motivates inter-disciplinary research across wireless communications, computing/machine learning, circuits/hardware, etc. This Special Issue is focused on the energy-efficient design of joint sensing, communication, computation, and control functionalities that collectively enable distributed AI in next-generation wireless networks. Topics of interest include, but are not limited to:

• Energy-efficient machine learning, communication, and control at the wireless edge
• Energy-efficient federated learning, including both supervised and reinforcement learning
• Signal processing techniques for energy optimization in distributed AI
• Novel PHY and MAC layer techniques to minimize energy consumption in distributed ML
• Networking design to enable energy-efficient large-scale distributed AI
• Green IoT for distributed AI
• Hardware design that reduces energy consumption for distributed AI
• Proof-of-concepts and experiments for energy-efficient distributed AI
• Energy-efficient distributed AI applications

Submission Guidelines

Authors need to follow the manuscript format and an allowable number of pages described at the IEEE TGCN Information for Authors page. To submit a manuscript for consideration for the special issue, please visit the journal submission website at Manuscript Central.

Important Dates

Submission Deadline: 30 September 2021 14 October 2021 (Extended Deadline)
First Review Due: 15 November 2021
Revision Due: 15 December 2021
Second Review Notification: 17 January 2022
Final Manuscript Due: 30 January 2022
Publication Date: March 2022 (Tentative)

Guest Editors