Cellular Licensed Band Sharing Technology among Mobile Operators: A Reinforcement Learning Perspective

Next-generation wireless networks will need to support of very high data rates and low–latency communications, which will require a new wireless radio technology paradigm. The growing number of mobile users is causing spectrum scarcity; and hence, an efficient spectrum utilization method is required. Conventional scheduling-based resource allocation scheme in wireless networks under limited resources is a challenging due to the complex network situations, dynamic network environment, and diverse needs for future networks. To overcome resource scarcity in mobile networks, spectrum sharing among multiple operators may be an efficient solution. Traditional methods of dynamic spectrum sharing are model-dependent, and they are not robust to the changing wireless environments. To enable low-latency communications for complex future wireless networks, efficient machine learning algorithms can be used across the wireless network infrastructure. Integrating machine learning for resource allocation can leverage intelligent and efficient mechanisms for dynamic wireless networks. To efficiently and intelligently utilize the scarce resources of dynamic networks, this paper proposes an efficient machine learning-based spectrum sharing method among multiple mobile network operators (MNOs). A mobile network operator uses the idle slots of the another operator and transmits the information efficiently. Using the neural network model, each MNO learns the spectrum utilization of other MNOs and selects the idle slots of other MNOs. Simulation results have been computed and compared with the conventional scheme where resources are not shared. These simulation results show that the proposed neural network model can efficiently learn the network quickly, and spectrum sharing can lead to improved network performance in terms of the delay, user-perceived throughput, resource usage, packet drop, and sum throughput of the network.