NTN Spectrum & Bands (L/S-band FR1, Ka-band FR2)
The FR1-NTN bands (n255/n256 in L/S-band) and FR2-NTN Ka-band, the UE (38.101-5) and satellite-access-node (38.108) RF requirements, and duplexing.
Which band an NTN link uses is decided by one question: is the terminal a bare handset or a dish? A phone with a tiny antenna needs low frequencies — L-band and S-band in FR1 — where the link budget is forgiving. A fixed VSAT with a real dish can afford Ka-band in FR2 and the bandwidth that comes with it. This page covers the NTN bands, the two RF specs, and why direct-to-handset NTN is not mmWave.
Introduction
Spectrum is where NTN's physics meets its product. The delay and Doppler of an orbit are fixed by geometry, but the band an operator picks is a design choice — and it is driven almost entirely by the terminal. A satellite serving smartphones and one serving fixed dishes want opposite ends of the spectrum, and 3GPP defines separate operating bands and separate RF specifications for each.
This matters at exactly the moment a network is designed or a device is certified. Choosing FR1-NTN (L/S-band) commits you to a forgiving link budget but narrow bandwidth and modest throughput; choosing FR2-NTN (Ka-band) unlocks wide bandwidth but only closes the link for a high-gain dish. The RF requirements — emission masks, sensitivity, polarisation — differ enough that the handset and the satellite radio each get their own document.
It is a conceptual page: it names the bands (n255, n256, Ka), the two RF specs (TS 38.101-5, TS 38.108), and the link-budget reasoning (free-space path loss, antenna gain, EIRP) that ties a terminal type to a frequency. The satellite-assistance broadcast that tells the UE which band and polarisation to use is ntn-sib19.
On this page
Why the terminal picks the band
In plain words: shouting to someone across a valley is like an NTN link. If all you have is your bare voice (a handset's tiny antenna), you shout in a low, booming tone that carries — high-pitched sound scatters and dies over distance. If you have a megaphone aimed straight at them (a VSAT dish), you can use a sharper, higher tone and pack far more words in, because the megaphone's focus makes up for it. The gear at your end decides which "pitch" (frequency band) actually reaches the other side.
The reason is the link budget. A satellite is hundreds to tens of thousands of kilometres away, so the free-space path loss is enormous — and it grows with frequency. A terminal has only two levers to fight that loss: transmit power and antenna gain. A handset has almost no gain (a small near-omnidirectional antenna) and limited power, so it must operate where the path loss is smallest — low frequency. A VSAT has a high-gain dish, so it can afford the extra loss at high frequency in exchange for the bandwidth that lives there.
So the terminal is the independent variable and the band is the consequence. Everything else on this page — the two RF specs, the specific bands, the FDD/polarisation choices — follows from matching a frequency to what the terminal's antenna and power budget can actually close.
Two RF specifications
Terrestrial NR splits its RF requirements between the UE (TS 38.101-1 for FR1, TS 38.101-2 for FR2) and the base station (TS 38.104). NTN mirrors that split with its own dedicated documents, because a satellite radio and a satellite-facing handset have requirements that differ enough to warrant separate specs.
TS 38.101-5 is the NTN-specific UE RF spec — separate from the terrestrial TS 38.101-1/2. It defines the transmit/receive requirements for a UE operating over satellite.
TS 38.108 defines the RF for the Satellite Access Node (the NTN gNB / on-board or ground radio), the counterpart to terrestrial TS 38.104.
NTN links run at very low SNR over enormous distances with circular polarisation and specific MSS bands, so the emission masks, sensitivity and band tables differ enough to need their own documents.
Remember the pair: TS 38.101-5 = NTN UE RF; TS 38.108 = NTN Satellite Access Node RF. If someone asks "which spec covers the NTN handset's RF," the answer is 38.101-5, not 38.101-1.
FR1-NTN — L-band and S-band for handsets
The whole point of direct-to-handset NTN is that an ordinary-looking phone gets service from space. That only works at low frequency, so FR1-NTN lives in L-band and S-band using Mobile Satellite Service (MSS) spectrum, with FDD duplexing.
| Band | Range | Rough frequency | Duplex | Target terminal |
|---|---|---|---|---|
n255 | L-band | ~1.5 / 1.6 GHz (MSS) | FDD | Handheld direct access |
n256 | S-band | ~2 GHz (MSS) | FDD | Handheld direct access |
Lower frequency buys two things a handset desperately needs: a friendlier free-space path loss and better behaviour with a small, near-omnidirectional antenna that cannot form a high-gain beam. That is why the headline Rel-17 goal — direct-to-cell for smartphones — sits in L/S-band and not up in mmWave.
These are also relatively narrow bands. MSS allocations at L- and S-band are measured in tens of megahertz, not the hundreds available at Ka-band, so per-user throughput over a handset link is modest. That is a deliberate trade: the priority for FR1-NTN is closing the link to a phone at all, and moving a useful but limited data rate, rather than maximising peak capacity. Coverage and reachability win over raw speed.
FR2-NTN — Ka-band for VSAT
When the terminal is a fixed installation with a directional antenna — a VSAT dish — the calculus flips. A dish delivers high gain, which pays for the harsher path loss at high frequency, and in return you get the wide bandwidths only available up there. FR2-NTN therefore uses Ka-band, roughly 20 GHz downlink and 30 GHz uplink, for high-throughput fixed and transportable terminals.
Ka-band FR2 NTN: ~20 GHz DL / ~30 GHz UL, aimed at VSAT and other directional-antenna terminals rather than bare handsets.
Far more bandwidth than L/S-band, so far higher data rates — practical only because the dish's gain closes the link at these frequencies.
The high-gain directional antenna compensates for the larger path loss and lets the terminal track the satellite; it also makes the (larger absolute) Ka-band Doppler manageable. See ntn-doppler.
Note the asymmetric duplex frequencies — ~20 GHz down, ~30 GHz up. Uplink sits higher partly because the satellite's receiver and the terminal's dish can tolerate the extra loss there, and it follows long-standing Ka-band satellite plans. The dish also has to physically track the satellite as it moves (especially for LEO), which a fixed handset antenna cannot do — another reason Ka-band is a VSAT story, not a handset one.
Why handsets go low-band, not mmWave
The instinct from terrestrial 5G is that mmWave means capacity. Over satellite that instinct is wrong for a phone, and it is worth being precise about why. The satellite is hundreds to tens of thousands of kilometres away, so the slant range is enormous and the free-space path loss is punishing. A terrestrial mmWave cell survives that geometry with big phased arrays at both ends; a satellite has limited EIRP and a handset has a tiny antenna that cannot form a high-gain beam. At mmWave the handset link simply would not close.
The physics is captured by the free-space path loss (FSPL): the fraction of power lost simply by the signal spreading out over distance, which rises with both distance and frequency.
Two terms fight the handset. The distance term is brutal on its own — a 600 km LEO link is ~154 dB of loss at 2 GHz, and a GEO link tens of dB worse. The frequency term then adds insult: moving from 2 GHz (S-band) up to 20 GHz (Ka) adds 20·log10(10) ≈ 20 dB of extra loss for the same distance, and 30 GHz uplink adds even more. A handset's small antenna and limited power have no way to recover 20+ dB, so it must stay low. A VSAT dish, with tens of dBi of gain, can — which is exactly the trade FR2-NTN makes.
Drop to L/S-band and two things improve at once: the free-space path loss is lower, and a small omnidirectional antenna radiates efficiently without needing to beamform. That combination is what makes direct-to-device feasible. Ka-band is reserved for terminals that do have a high-gain dish and can therefore trade frequency for bandwidth.
Duplexing, polarisation and the low-SNR reality
NTN links are typically FDD — separate uplink and downlink frequencies — which suits the long, symmetric round trips better than trying to time-share one carrier across hundreds of milliseconds of delay. Satellite links also commonly use circular polarisation rather than the linear polarisation of terrestrial cells, because a spinning or tumbling geometry and Faraday rotation through the ionosphere make a fixed linear orientation unreliable.
The TDD-versus-FDD point is worth making concrete: TDD works by rapidly switching one carrier between transmit and receive, but that scheme assumes the round-trip is short compared with the switching period. With NTN round-trips in the tens to hundreds of milliseconds, a TDD guard period would have to be enormous and most of the frame would be wasted waiting for the far end — so paired FDD spectrum is the natural fit.
Because polarisation is now a live parameter, the network tells the UE which polarisation to use on each direction: ntn-PolarizationDL and ntn-PolarizationUL are signalled in SIB19. The full contents of that satellite-assistance broadcast are covered in ntn-sib19.
Low-SNR design: a satellite has limited EIRP and the slant range is huge, so NTN routinely operates at very low SNR. Coverage-enhancement techniques — repetition, robust low-order MCS, strong coding — carry the link. That is why HARQ and coverage tooling matter so much here; see ntn-harq and ntn-evolution.
TN ↔ NTN: terrestrial NR spreads across FR1 and mmWave FR2, uses TDD heavily (especially in the mid-band), assumes linear polarisation, and splits RF requirements across TS 38.101-1/2 (UE) and TS 38.104 (base station). NTN inverts most of that: handsets sit in low L/S-band MSS spectrum (n255/n256), VSATs in Ka-band, duplexing is FDD, polarisation is circular and signalled per-direction, and RF gets its own pair of specs — TS 38.101-5 (UE) and TS 38.108 (Satellite Access Node). The waveform is the same NR; the band plan and RF envelope are re-drawn for space.
⚠ Common pitfalls / gotchas
- Assuming NTN uses the terrestrial UE RF spec. The NTN handset is certified against
TS 38.101-5, notTS 38.101-1/2; the satellite radio againstTS 38.108, notTS 38.104. - Expecting mmWave-style throughput from a direct-to-handset link. FR1-NTN is narrow MSS spectrum (tens of MHz) chosen for reachability, not peak rate — Ka-band bandwidth is a VSAT-only benefit.
- Ignoring polarisation. NTN uses circular polarisation signalled by
ntn-PolarizationDL/ntn-PolarizationULinSIB19; treating it like terrestrial linear polarisation costs a polarisation-mismatch loss. - Reaching for TDD. The huge round-trip makes TDD guard periods impractical, so NTN bands are paired FDD.
- Forgetting the frequency term in FSPL. The ~20 dB penalty from 2 GHz to 20 GHz is the whole reason a handset cannot use Ka-band — distance alone is not the only obstacle.
Summary
The terminal decides the band, and the link budget is why. A bare handset has a tiny antenna and limited power, so it must live where free-space path loss is smallest — FR1-NTN in L/S-band MSS spectrum (n255 ~1.5/1.6 GHz, n256 ~2 GHz), FDD, narrow bandwidth, coverage over peak rate. A VSAT with a high-gain dish can absorb the ~20 dB extra FSPL at high frequency in return for wide bandwidth — FR2-NTN in Ka-band (~20 GHz DL / ~30 GHz UL). That is why direct-to-cell is low-band, not mmWave: the handset simply cannot close a Ka-band link.
NTN carries its own RF specifications — TS 38.101-5 for the UE and TS 38.108 for the Satellite Access Node — because the low-SNR, long-distance, circularly polarised MSS environment needs different masks, sensitivity and band tables than terrestrial NR. Duplexing is FDD (TDD guard periods would be impractical over hundreds of ms), polarisation is circular and signalled per-direction via ntn-PolarizationDL/ntn-PolarizationUL in SIB19, and the whole link leans on coverage-enhancement coding. Next, see how the UE learns all of this from the satellite-assistance broadcast in ntn-sib19.
Q. Why is direct-to-handset NTN in L/S-band rather than mmWave?
A. The slant range to the satellite makes free-space path loss enormous, the satellite has limited EIRP, and a handset's tiny omnidirectional antenna cannot form a high-gain beam. Low frequency reduces path loss and radiates efficiently from a small antenna, so the link closes. mmWave is only viable for VSAT terminals with high-gain dishes — that is what Ka-band FR2 targets.
Q. What is the difference between TS 38.101-5 and TS 38.108?
TS 38.101-5 is the NTN-specific UE RF specification (the satellite counterpart to terrestrial 38.101-1/2). TS 38.108 is the RF specification for the Satellite Access Node — the NTN gNB radio — the counterpart to terrestrial 38.104.
Q. Which NTN operating bands are defined for FR1, and what duplex mode?
n255 in L-band (~1.5/1.6 GHz MSS) and n256 in S-band (~2 GHz MSS), both FDD, both aimed at direct handheld access.
Where to go next
Band and polarisation are announced to the UE by the satellite, and the low-SNR reality drives the coverage and reliability toolkit. Follow the assistance broadcast and the reliability chain.