6G Vision: An AI-Driven Decentralized Network and Service Architecture Xiuquan Qiao State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications Yakun Huang Beijing University of Posts and Telecommunications Schahram Dustdar Vienna University of Technology Junliang Chen Beijing University of Posts and Telecommunications Abstract—Recently, following the rapid commercial deployment of 5G networks, next-generation mobile communication technology (6G) has been attracting increasing attention from global researchers and engineers. 6G is envisioned as a distributed, decentralized, and intelligent innovative network. However, existing application provisioning is still based on a centralized service architecture, ubiquitous edge computing, and decentralized AI technologies have not been fully exploited. In this article, we analyze the problems faced by existing centralized service provisioning architecture, and propose design principles for a decentralized network and service architecture for a future 6G network. Finally, we discuss several open research problems to inspire readers to address these. & DUE TO THE large number of commercial appli- cations of 5G networks worldwide, potential 6G technologies are attracting attention from both academia and industry. Although 5G has achieved significant improvements in terms of communi- cation performance, it remains difficult to meet demand for more intelligent communication in terms of information speed, multidomain cover- age, artificial intelligence (AI), and security. 1 Recently, several governments have launched 6G projects to explore the requirements and key technologies of the next-generation mobile com- munication network. However, existing visions and discussions of 6G mainly focus on innovative wireless communication technologies, mobile Digital Object Identifier 10.1109/MIC.2020.2987738 Date of current version 9 September 2020. Department: Internet of Things, People, and Processes Department: Internet of Things, People, and Processes Editor: Schahram Dustdar, dustdar@dsg.tuwien.ac.at July/August 2020 Published by the IEEE Computer Society 1089-7801 ß 2020 IEEE 33 Authorized licensed use limited to: Universitaetsbibliothek der TU Wien. Downloaded on September 16,2020 at 06:54:24 UTC from IEEE Xplore. Restrictions apply.