An Intelligent Hybrid Radio Access Technology Selection Algorithm for 5G-Satellite Network

Cellular networks are anticipated to handle a significant increase in data traffic, as well as a huge number of devices and new use cases in the very near future; therefore, future 5G networks are being built with heterogeneity in mind. The user's device will be able to connect to the network using 5G and other technologies such as LTE, WiMAX, WiFi, Satellite, etc. Satellite Communication (SatCOM), on the other hand, has been proposed as a network of choice to supplement the 5G use cases. In most of 5G's use cases, satellite communication would provide a complimentary service for global coverage, critical communication, broadcast/multicast provision, and multimedia traffic growth. Hence, for these connections to operate effectively, a smooth transmission between 5G and satellite is needed. However, in a heterogeneous wireless network, a major issue is determining how to assign a user to the best Radio Access Technology (RAT). An intelligent hybrid RAT selection algorithm based on machine learning is proposed in this paper. The User Terminal is trained to select the necessary RAT to admit its call using the Actor-Critic Reinforcement Learning Algorithm. Based on the simulation results, the proposed algorithm has the ability to learn in a competent way to successfully manage resource management autonomously and therefore increases the user's quality of service (QoS). The algorithm also showed better expected total reward, lower call blocking probability and higher throughput when compared with a Markov Reward-Based Greedy Algorithm.