Wi-Fi aware radio resource management in 5G NR-U: a learning-based coexistence scheme for C-V2X

Abstract

With the increasing number of internet of vehicles devices and the exceptional growth in data traffic, the licensed spectrum is faced with limitations in meeting the growing demand for cellular vehicle-to-everything (C-V2X) applications. Opportunistic utilization of unlicensed bands is regarded as a solution to this issue. However, using unlicensed bands by cellular technologies poses challenges for coexisting with other unlicensed systems. This research examines how 5G new radio operating in the unlicensed band (5G NR-U) can coexist with Wi-Fi systems. It assumes that 5G NR-U exploits the duty cycle method for managing coexistence. An optimization problem is established that exploits the estimated load of Wi-Fi systems to enhance the total throughput of the cellular network while considering the rate constraint of Wi-Fi users. In most coexistence schemes, the cellular system exploits knowledge of the Wi-Fi traffic through a given signaling channel. However, this signaling channel is not always applicable in practice. As a solution, this paper proposes an approach that exploits a federated convolutional neural network (CNN) to gauge the intensity of Wi-Fi traffic by analyzing unlicensed channel activity. Based on CNN’s prediction, a Q-learning based algorithm is then developed to solve the resource allocation problem and adjust the parameters of the duty cycle based on the estimated Wi-Fi load and C-V2X network status. Simulation results demonstrate that even without signaling exchanges, the suggested approach enhances the throughput of the cellular network by about 35% on average in scenarios with medium traffic load compared to the previous method while the required rates of Wi-Fi users are not considerably violated. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2025.