Improving Capacity for Real-Time Traffic in 5G NTN: Common and Unique Effects versus Terrestrial Systems

5G Non-Terrestrial Networks (NTN) is a new technology recently developed by 3GPP to achieve almost global coverage by deploying 5G base stations at high altitude platforms or satellites. From various deployment scenarios considered by 3GPP, the usage of Low Earth Orbit (LEO) satellites is the most promising way for providing a large variety of broadband applications (web, voice, video) even for handheld users. Existing research on 5G NTN LEO systems mostly focuses on evaluation and optimization of the network performance for the basic mobile broadband traffic, which is tolerable to high delay relevant to satellite links. However, end-users are also running real-time applications that impose strict Quality of Service (QoS) requirements on packet delivery time and reliability. This paper provides a comprehensive analysis of 5G NTN systems’ performance with real-time traffic. Using our own developed model of the 5G NTN systems in the ns-3 simulator, we analyze how the network capacity depends on the considered scenario, QoS requirements, and the link-layer algorithms used at the gNB. Based on the results, we explicitly differentiate which effects are common to 5G NTN and 5G terrestrial systems and which effects are unique to 5G NTN and, thus, require new solutions to be developed. Also, we provide recommendations for selecting algorithms and tuning their parameters to achieve a high capacity of the 5G NTN system for real-time traffic.