Learning-Assisted Clustered Access of 5G/B5G Networks to Unlicensed Spectrum
Qimei Cui, Wei Ni, Shenghong Li, Borui Zhao, Ren Ping Liu, Ping Zhang
Abstract
License-assisted access (LAA) to unlicensed spectrum is a potential solution to improve the resource availability and system scalability of 5G/ B5G networks. Challenges arise from coexistence between LAA and incumbent systems, especially the ubiquitous IEEE 802.11 WiFi systems. This article demonstrates that the coexistence can be substantially improved by leveraging learning-based clustering of small base stations (SBSs), referred to as learning-assisted clustered access (LACA), and improving the interoperability between licensed and unlicensed access. Fast signaling and a centralized control plane of licensed access can facilitate clustering SBSs for coordinated access to the unlicensed spectrum, hence reducing the number of parties contending for access and alleviating contention. Appropriate clustering of SBSs is important to the efficiency of LACA in 5G/B5G networks. LACA can quickly converge with strong locality, facilitating the coordination of the SBSs, for example, cooperatively connecting multiple users and conducting beamforming. Analytic evaluation and numerical tests confirm the improved coexistence through enlarged LAA-WiFi capacity regions, as well as reduced transmission delays.