Logo image
Construction of 6G Omni-Scenario Knowledge Base for On-Demand Service Using Channel Knowledge Map
Conference poster   Open access

Construction of 6G Omni-Scenario Knowledge Base for On-Demand Service Using Channel Knowledge Map

Zifan Sha, Changle Li, Wenwei Yue, Xiaojian Yang, Yuchuan Fu, Luqiao Wang, Nan Cheng, Mahdi Boloursaz Mashhadi Dr and Zhili Sun
AAAI 2026 Workshop on Machine Learning for Wireless Communication and Networks (ML4Wireless)
Association for the Advancement of Artificial Intelligence
AAAI 2026 Workshop on ML4Wireless (27/01/2026)
40th Annual AAAI Conference on Artificial Intelligence (AAAI-26) (Singapore, 20/01/2026–27/01/2026)
26/01/2026

Abstract

6G omni-scenario on-demand service channel knowledge map knowledge base construction
With the rapid evolution of 6G networks, the need for on-demand service across diverse scenarios continues to grow. However, a key challenge lies in the lack of a structured knowledge infrastructure that can possess the capability for standardized representation and accurately reflect the comprehensive 6G scenarios. To address this, we propose OSOD 6G, a 6G omni-scenario knowledge base empowered by Channel Knowledge Map (CKM), designed to capture and infer the essential features of wireless channels under varying environments. Firstly, by elucidating the intrinsic relationships among subject, environment, requirement, and knowledge through an ontological framework, OSOD 6G covers seven representative scenarios including multi-stage urban areas , plain, mountain, and ocean. Secondly, we deploy heterogeneous transmitters such as omnidirectional base stations, directional base stations, and satellites across distinct urban development stages, and densely position receivers both in aerial and terrestrial domains. Then, CKM inference is used to extract channel data, which is then filtered, struc-tured, and indexed under a unified schema. The resulting dataset includes 759,590 structured entries, enabling efficient scenario-aware inference. Finally, simulation results reveal significant channel characteristic variations across different deployment configurations and environmental types, underscoring the necessity for service customization. Moreover, in handling large-scale CKM data, the XGBoost model demonstrates superior fitting accuracy compared with the baseline, achieving an R-squared of 94.94% and a root mean square error (RMSE) of 0.8394.
pdf
10_Construction_of_6G_Omni_Sce2.50 MBDownloadView
Author's Accepted Manuscript Open Access CC BY V4.0
url
https://sites.google.com/view/ml4wirelessaaai26View
Event Website Workshop website
url
https://openreview.net/group?id=AAAI.org/2026/Workshop/ML4Wireless#tab-accept-posterView
Workshop proceedings

Metrics

1 File views/ downloads
2 Record Views

Details

Logo image

Usage Policy