ML-Based Dynamic Network Switching Framework for Nonterrestrial Networks in 5G and Beyond

Nonterrestrial networks (NTN) have gained significant ground with 5G and beyond wireless communications systems. The utilization of NTN presents strong benefits under various scenarios in terms of coverage, scalability, and ubiquity compared to the terrestrial networks (TN). In addition, different NTN platforms may have several advantages over each other. In this study, a dynamic network switching framework for NTN systems is developed to decide on usage of the NTN platform in a region. Under this framework, the first proposed method is designed to determine the necessity of NTN usage for a specific region. The second proposed method decides which NTN platform is the most reasonable for meeting the NTN usage necessities of the users in the same region. By consecutively employing these two methods, cellular communications operators can automatically manage heterogeneous NTN that are integrated into TN infrastructure while increasing the overall network efficiency. Synthetic datasets are generated by using environmental information and communications requirements. Then, machine learning (ML) algorithms are utilized to form the proposed dynamic network switching framework. A discussion regarding the performance evaluation of the preferred ML algorithms is provided. Moreover, a use case scenario is given to demonstrate the merits of the proposed framework.