The frequency spectrum availability is the most pertinent characteristic of NTN communications. Since satellites are not restricted to one country or region, an international harmonization of frequencies is essential for global satellite communications. Currently, several frequency ranges are being discussed for NTN. The current Frequency Range 1 (FR1) bands agreed up by 3GPP for NTN are the S-Band frequency range of 1980 to 2010 MHz in the uplink (UL) and 2170 to 2200 MHz in the downlink (DL), which is band n256 and the L-Band frequencies 1525 to 1559 MHz DL together with 1626.5 to 1660.5 MHz for the UL, which is band n255.

In the longer term, 3GPP is discussing NR-NTN above 10 GHz. Ka-Band is the highest priority band with a DL of 17.7 to 20.2 GHz and a UL of 27.5 to 30 GHz. Ku-Band with a DL of 10.7 to 12.75 GHz and a UL of 12.75 to 13.25 GHz and 13.75 to 14.5 GHz is also being envisaged. While these frequency bands are in use for current satellite communications, they present challenges for 5G implementations. Some of these bands fall into the spectrum gap between the 5G FR1 and FR2 bands. In addition, NTN frequencies will use frequency-division duplexing, due to the long round-trip time, as opposed to the time-division duplexing scheme that is widely used by 5G NR.

Like terrestrial communications, coexistence is relevant for NTN. A satellite cell or beam coverage area is large and often exceeds country and terrestrial cell borders. Deployments must address supplementary satellite coverage where terrestrial and non-terrestrial networks share the spectrum or different spectrum bands.


The following architectures are relevant for current and future NTN and satellite constellations:

Low earth orbit (LEO): Satellites with an altitude between 500 km and 2000 km have a shorter round-trip time (RTT), which is typically less than 30 ms. The size of a LEO satellite is also assumed to be small, with a diameter typically < 1 m and may even be in the range of a dozen centimeters for a nanosatellite, with a weight below 500 kg. The assumption is that NTN uses a beamforming mechanism at the satellite station. The typical beam footprint of a LEO satellite ranges between 100 km and 1000 km.

Medium earth orbit (MEO): Satellites travel at a velocity of about 13,800 km/h and have an orbital period of 6 to 12 hours. The beam footprint is like a LEO constellation.

Geostationary earth orbit (GEO): Satellites operate above the equator at an altitude of 35,786 km resulting in a notional station keeping its position fixed in terms of elevation and azimuth angle with respect to a given Earth point. The beam footprint ranges from about 200 km for narrow beams, up to 4000 km in the case of large beams. Due to the larger orbit radius distance, the RTT of a GEO satellite is about 544 ms.

High altitude platform systems (HAPS) or high altitude IMT base stations (HIBS): This category includes airborne objects such as airplanes, balloons, helicopters and drones (UAVs). They operate very flexibly at altitudes from several hundred meters up to about 15 km and have beam footprints with diameters of just a few kilometers up to 100 km, on average. Operators may use HAPS/HIBS to provide additional capacity in a specific region, making dynamic deployment an advantage. A disadvantage of this architecture is the smaller coverage area. Due to the shorter distances, the RTT performance is competitive with terrestrial networks. The HAPS and HIBS use cases are differentiated by spectrum usage and use cases. HAPS networks currently focus on FSS only, due to regulations, while HIBS networks may provide MSS.


The satellite covers a geographical area by forming a beam. That beam footprint is either static or moving with respect to Earth. NTN architectures need radio access from the terrestrial terminal or UE to the satellite, which is referred to as the service link. To complete this overall link, the satellite needs to be connected to a terrestrial gateway, referred to as the feeder link. LEO and GEO satellite constellations have a known or predictable trajectory, which facilitates the routing of the connection to the ground station. To facilitate NTN-capable RAN deployment, the 3GPP is discussing transparent mode and regenerative mode architectures. Release 17 deals primarily with the transparent mode architecture.

Transparent NTN NG-RAN architectures behave like a repeater or bent pipe in space. This architecture disaggregates the terrestrial base station into the satellite components, ground gateway and terrestrial gNB functions. The satellite functions implement RF filtering, frequency conversion, RF amplification, RF transmission and reception in the uplink as well as the downlink direction. The pivotal characteristic of this architecture is that the waveform is repeated between the service link and feeder link by an unchanged payload. The carrier undergoes a frequency change for a variety of reasons, including avoiding interference between the service and feeder links.

This architecture is independent of the radio waveform, so any changes do not require modifications in the spaceborne station. Disadvantages of this architecture include noise amplification as the satellite may not perform any channel equalization or noise cancellation, vulnerability against jamming attacks, longer overall RTT with two satellite-Earth links and the lack of inter-satellite link (ISL) connections for traffic steering.

The connection between the UE and the terrestrial gNB not only includes the service link and the feeder link but several ISLs in between these two links are possible in future extensions. 3GPP TR 38.821 states that the regenerative payload is required for the first ISL implementations. Figure 3 shows a representation of this transparent NTN NR-RAN network.

Figure 3

Figure 3 Transparent payload NTN NG-RAN architecture. Source: Rohde & Schwarz.

Future NTN deployments that Release 19 will define will include regenerative mode architectures. The major difference from the transparent payload architecture is that gNB functions are incorporated into the satellite in regenerative mode architectures, creating faster scheduling decisions and shorter RTT. The regenerative architecture model raises satellite hardware complexity and computing power and may also incorporate multi-access edge computing (MEC) functionalities to reduce the RTT.


The distance between terrestrial UE and spaceborne stations impacts the link budget or high path attenuation, but simulation results show that the signal-to-noise ratio conditions permit communication. More critical is the long time delay or RTT, which also depends on the time and elevation angle. Satellite velocity causes a frequency carrier deviation or Doppler shift, which creates a paradigm change compared to terrestrial networks where the base station is stationary. Ionospheric radio wave propagation is also responsible for waveform polarization rotation, known as Faraday rotation.

Path Attenuation

The distance between the UE and the satellite creates high path attenuation. 3GPP discussed several link budgets and carried out studies with diverse parameters and simulation results, which are shown in TR 36.763 and TR 38.811. As the antenna technology evolves, the objective is to lessen the path loss challenge with highly directive antennas that increase antenna gain. The composite path loss is based on the basic path loss, which is mainly the free-space path loss (FSPL), attenuation due to atmospheric gases, attenuation due to atmospheric scintillation and building entry path loss. Typical assumptions are FSPL values of -160 dB for LEO and -190 dB for GEO and it is assumed that the UE RX sensitivity will be better than terrestrial networks.

RTT and Differential Time Delay

The large distance between the terrestrial UE and the satellite creates a long RTT. This creates a long latency period and it poses a challenge to low latency communication applications planned with NR-NTN. Typical one-way latency values range from 30 to 40 ms in LEO constellations and up to 544 ms in GEO constellations.

A detailed analysis of the RTT and latency aspects of satellite networks identifies two challenges. The first concerns the differential delay between the NTN gNB and all the UE in a beam footprint coverage area. The second is the time-varying latency and RTT during the entire connection period due to the nature of an elliptical flight orbit and the changing distance between the UE and the satellite. The first challenge is caused by the elliptical shape of the beam footprint and how the size of the ellipse depends on and changes with the elevation angle. This means that the satellite experiences different propagation times among the UE within the beam footprint. The second challenge is caused by the UE experiencing an RTT that varies in response to the satellite’s orbital trajectory. When the satellite appears at the horizon, just above the minimum elevation angle, the distance between the UE and the gNB is the longest. This creates the largest value of RTT, but this will change with the elevation angle. This impacts the buffer management of the MAC layer and HARQ operation.

Doppler Frequency Shift

One of the most serious challenges to creating NTN connections with good quality of experience is carrier frequency deviation or Doppler shift. A moving base station or satellite, in combination with a UE potentially moving, causes a time-variant Doppler shift across the connection time. This Doppler shift depends on the relative velocity between the UE and the satellite, the carrier frequency and the angle between the velocity vector and the signal propagation direction.

Faraday Rotation in NTN and Polarization Aspects

Faraday rotation is caused by the structure of the atmosphere and it is indicated by the total amount of electrons. Faraday rotation describes the rotation of the polarization resulting from the interaction of the electromagnetic wave with the ionized medium in the Earth’s magnetic field along the path. This is described in TR 38.811. Circular polarization methods may counteract this effect, but this method requires the UE to apply the same circular polarization or tolerate a 3 dB polarization loss in addition to the FSPL.


The goal of 3GPP is to enable 5G NTN, satellite-based communications with the lowest impact on 5G. As communications evolve, the architecture of wireless networks must also evolve away from the cellular network concepts of previous wireless generations.2 The anticipation is that 6G will consist of multiple dynamic and intelligent nodes with onboard computing power and MEC functionality that are interconnected and may be moving relative to each other. The terms interworking, integration and unification describe the evolution path from legacy satellite and cellular technologies to 5G NTN and eventually, 6G.

New research areas will enable an evolution toward organic networks. These organic networks will incorporate cell birth and death behavior, vagabonding network components and intelligent traffic management. Incorporating NTN into the 5G ecosystem with Release 17 signals the advent of a new technology evolution fostering and driving the worldwide proliferation of wireless communications systems.


  1. R. Stuhlfauth, “5G NTN Takes Flight: Technical Overview of 5G Non-Terrestrial Networks,” Rohde & Schwarz, Whitepaper, https://www.rohde-schwarz.com/solutions/test-and-measurement/aerospace-defense/satellite-test/white-paper-5g-ntn-takes-flight-technical-overview-of-5g-non-terrestrial-networks_255919.html.
  2. “5G & Non-terrestrial Networks,” 5G Americas, Whitepaper, February 2022.