5G NR Non-Terrestrial Networks: Open Challenges for Full-Stack Protocol Design

TL;DR

The paper tackles the challenge of integrating Non-Terrestrial Networks into 5G NR by surveying the Rel.17–Rel.20 NTN landscape and documenting open full-stack protocol design questions. It introduces an open-source, end-to-end ns3-NTN simulator to quantify the effects of Doppler, synchronization, framing, and mobility on NTN performance. Key contributions include a structured treatment of synchronization, resource allocation, duplexing, HARQ, handover, routing, and transport, complemented by system-level results illustrating throughput and latency implications across NTN scenarios. The work informs standardization activities and provides a practical pathway toward robust NTN deployment in current and future generations of wireless networks.

Abstract

As 5th generation (5G) networks continue to evolve, there is a growing interest toward the integration of Terrestrial Networks (TNs) and Non-Terrestrial Networks (NTNs). Specifically, NTNs leverage space/air base stations such as satellites, High Altitude Platforms (HAPs), and Unmanned Aerial Vehicles (UAVs) for expanding wireless coverage to underserved rural/remote areas, supporting emergency communications, and offloading traffic in highly congested urban environments. In this paper we focus on the 3GPP 5G NR-NTN standard in the context of satellite communication networks, and highlight critical challenges that must be addressed for proper full-stack protocol design, with considerations related to the PHY, MAC, and higher layers. We also present simulation results in ns-3 to demonstrate the impact of some of these challenges on the network, as an initial step toward more advanced standardization activities on 3GPP 5G NR-NTN.

Figures (4)

• Figure 1: Impact of the differential delay among different UEs on the throughput. We consider $N_u=4$ communicating in S band and generating UL UDP data with a source rate of 25 Mbps each, so the cumulative cell data rate is 100 Mbps. We compare the case in which all UEs are deployed around the cell center (plain bars) vs. the case in which they are uniformly distributed across the cell (striped bars).
• Figure 2: Application latency as a function of $M$ (the number of consecutive DL slots) and $N$ (the number of GP slots where no transmissions can be scheduled). We consider a DL UDP flow with a source rate of 10 Mbps, and an uplink feedback flow with a source rate of 5 Kbps, between a terrestrial UE and a LEO satellite at an altitude of 600 km operating in the Ka band with numerology 3.
• Figure 3: Application throughput in the S band (at the cell center and at the cell edge) as a function of the satellite altitude, for different HARQ configurations. We change the number of HARQ processes, $n$, vs. a benchmark scheme where HARQ is disabled. We consider $N_u = 4$ UEs generating UL UDP data with a source rate of 25 Mbps, so the cumulative cell data rate is 100 Mbps.
• Figure 4: Average DL throughput (top) and latency (bottom) with UDP and TCP Cubic vs. the altitude of the satellite, $h$. The application generates data at a source rate of 10 Mbps, and both the transmitter (satellite-gNB) and the receiver (UE) are equipped with antennas to operate in good SNR regimes. The operating frequency is in the Ka band.