Map-Based Millimeter-Wave Channel Models: An Overview, Data for B5G Evaluation and Machine Learning

• Published in: IEEE wireless communications Vol. 27; no. 4; pp. 54 - 62
• Main Authors: Lim, Yeon-Geun, Cho, Yae Jee, Sim, Min Soo, Kim, Younsun, Chae, Chan-Byoung, Valenzuela, Reinaldo A.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.08.2020; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: 3GPP; 5G mobile communication; Algorithms; Channel models; Communications systems; Delays; Evaluation; Layout; Machine learning; Machine learning algorithms; Mathematical models; Millimeter wave communication; Millimeter waves; Parameters; Ray tracing; Researchers; Stochastic processes
• ISSN: 1536-1284; 1558-0687
• DOI: 10.1109/MWC.001.1900315

Within the mm-Wave range of the B5G communication systems, there will appear many types of applications with different link types. In discussions on how to cover the various channel modeling requirements of such applications, some researchers have suggested map-based channel models that are based on a ray-tracing algorithm. This article thus begins with an overview of mapbased mm-Wave channel models. The overview includes available modeling requirements with map-based channel models and the categorization of map-based channel parameters. We then explain why map-based channel models are necessary for researchers trying to evaluate novel technologies in the mm-Wave range. They are particularly necessary when the technologies operate with a new link type or when channel behaviors of a user exhibit user-specific characteristics. Next, we share the measurement data and the map-based channel parameters, which can model new link types and have user-specific characteristics. Finally, as a use case of the proposed channel model, we evaluate a machine-learning-based beam-selection algorithm that exploits power delay profiles through the shared database and a geometry-based stochastic channel model (GSCM). The delay profiles of the shared database are user-specific so they are highly correlated with angular parameters. Numerical results show that the algorithm can be more accurately evaluated through the shared database than through a GSCM. We expect that our overview and sharing the database will enable researchers to readily design a map-based channel model and evaluate B5G and machine-learning technologies.