Will satellites interfere with 5G? What the engineering studies submitted to 3GPP show
The FCC votes on 29 October on a proposal to open 482 MHz more spectrum to satellite service for ordinary phones. AT&T, T-Mobile and Verizon want the 2.5 and 3.45 GHz bands left out. SatNews summarized their trade group's warning: satellite operations "could degrade commercial mobile networks that support hundreds of millions of consumers." We collected the numbers from the engineering studies companies have submitted to 3GPP. Here is what they show.
Edited by Chris AckerPublished 2026-10-28, before the voteFCC open meeting 2026-10-29, 10:30 ET
The carriers' warning is about 2.5 GHz and 3.45 GHz, where their 5G networks use TDD: each tower talks and listens on the same frequency, in turns. None of the engineering studies we found covers either band, or a satellite sharing the same channel as a TDD network. Every 2026 result below is for a satellite one channel away from a 5G network near 2 GHz. 3GPP's other TDD cases are also one channel away: n34 near 2 GHz, and satellite bands at 11 to 27 GHz served by satellite dishes, not phones. These numbers show how well 3GPP's current rules work in that easier case. They do not tell you what would happen at 2.5 or 3.45 GHz.
Checked: TR 38.863 V20.0.0, clauses 6.1, 6a.1 and 6c.1; 15 RAN4 documents from RAN4 #120 and #120-bis (12 contributions and 2 moderator summaries under agenda 8.4.4, and 1 contribution under 8.4.3.2); RAN plenary documents from RAN #111 to #113; and a search of every TDoc in Scout's index since January 2025
Five findings
0 of the 7 companies whose 2026 results we read studied 2.5 GHz or 3.45 GHz. All studied a satellite one channel away from a 5G network near 2 GHz.
On whether 3GPP's current 24 dB satellite leakage limit protects a nearby 5G phone, 2 companies say yes or right at the line (MediaTek, Amazon), 2 say it does not or may not (AST SpaceMobile, Huawei), and 3 give numbers without a verdict (Qualcomm, LG, vivo).
The largest harm reported is AST SpaceMobile's: a city 5G user next to a 12,672-satellite constellation loses 6.9% of throughput at today's limit, above the 5% target these studies use, and 23% at a higher satellite power estimate.
How the 5G network is modeled can decide the answer: run with 3GPP's Release 17 network model, the same AST case shows essentially no loss.
Studies disagree on whether more satellites matter: LG found 645 visible satellites add 0.5 dB to its one-satellite result, while AST finds the needed protection rises steadily with constellation size.
Sources: the TDocs in the table below
The seven studies, side by side
These are simulations and analyses that companies submitted to 3GPP's radio working group, RAN4, not lab measurements. Each company chose its own assumptions and has a business stake in the outcome, so the table names who submitted each study: satellite operators, chip makers, phone makers and a network vendor.
Who ran it
What interferes with what
Key assumption
Result in plain terms
3GPP Rel-17 study, agreed by 3GPPTR 38.863, clause 6.5.1
Satellite downlink into the downlink of a 5G network one channel over, 2 GHz
One satellite at 600 km
Needs 23.81 dB of leakage protection (ACLR); 3GPP set the rule at 24 dB
ACLR, leakage protection: how much weaker a transmitter's spill into the next channel is than its own signal, in dB. More is better. 3GPP's limit for low orbit satellites is 24 dB.
ACIR, combined protection: how much a neighboring signal is weakened before it hits your receiver, counting both the sender's leakage and the receiver's filter, in dB. More is better.
Throughput loss: how much data speed the 5G network loses to the satellite. These studies treat 5% as the limit.
I/N: interference compared with the receiver's own noise, in dB. −6 dB means interference a quarter as strong as the noise.
Loss of sensitivity (desensitization): how much stronger a signal must be for the phone to still hear it. 3 dB means interference as strong as the phone's own noise.
dBW/m²/MHz: power arriving on each square meter of ground. −110 is ten times stronger than −120.
Qualcomm's first proposal reads: "Retain the Release 17 coexistence methodology". It asks that AST's aggregate method be calibrated across companies before it is used to change any requirement. Amazon argues the realistic count is 1 to 5 effective interferers at a phone, "not tens or hundreds", because operators keep their own beams on one frequency apart, and that each operator's filed ground limit caps total power.
R4-2621312 (Qualcomm), Proposals 1 and 2; R4-2620834 (Amazon), Observations 1, 3 and 4
Count every satellite and the network edge
AST says the Rel-17 model "is therefore not a conservative basis for this path and should not be used for it." Huawei says the one-satellite assumption of Rel-17 "may be no longer appropriate". Verizon says the Rel-17 model of a busy network hides interference that matters at the edge of coverage.
A satellite in the same channel as a TDD network at 2.5 GHz (n41) or 3.45 GHz (n77).
Satellite downlink landing on a 5G tower while it listens for phones, the problem the FCC raises for TDD bands, in the same channel at 2.5 or 3.45 GHz. The 11 to 27 GHz studies cover this only one channel away.
Direct-to-device phones transmitting near 5G towers in the same channel. The Rel-17 study looked only at neighboring channels, and the satellite-to-5G task RAN set for RAN4 in 2026 and the ITU's WRC-27 studies are downlink only.
Sources: TR 38.863 V20.0.0, clause 6.1, Table 6.1-2, and clauses 6a.1 and 6c.1 (Tables 6a.1-3 and 6c.1-3); RP-261531 (RAN #112 conclusion); FCC draft NPRM, paragraphs 75 and 121
TS 38.300In the specification
3GPP's satellite 5G standard is FDD only today
The Release 19 specification states: "In this release, NTN is only applicable to FDD system." The one exception is NB-IoT, a low-rate version for connected devices, which Release 19 extended to TDD for Iridium's L band. The satellite-to-5G task RAN set for RAN4 at RAN #112 covers one case, at 2.1 GHz, where both systems are FDD. In September, RAN did not take up the operators' proposals for new satellite bands; the moderator wrote that "companies remain divided".
CTIA and the three national carriers asked the FCC to remove both TDD bands, which together are 290 MHz, or 60.2% of the 482 MHz. T-Mobile's own filings, as reported, say up to 96% of a cell site's uplink coverage area could be lost where satellite coverage is present. In 3GPP, T-Mobile leads a proposal for satellite service in the PCS band, which is FDD, with SpaceX among the co-signers.
Sources: ECFS 26110083526 (CTIA, October 22), 26110083200 and 26110083663 (T-Mobile, October 22 and 23, 2026), filing records only, figures as reported by DCD, October 27, 2026; band edges from the FCC draft, paragraphs 29 and 30; RP-261832; PCS is n25, FDD, in TS 38.101-1 V20.1.0, clause 5.2
Satellite operators
SpaceX wants band work now; AST wants a study first
SpaceX wrote that "3GPP is respectfully urged to approve the work item proposals prepared by Operators", citing requests from the Australian, Canadian and US regulators. AST SpaceMobile proposed that no such work item be approved "before an independent coexistence Study Item for the relevant scenarios is completed." SpaceX and Amazon filed on the FCC draft on 24 October; we found no AST SpaceMobile or Globalstar filing on it.
Sources:RP-262009 (SpaceX), Proposal 1; RP-262096 (AST SpaceMobile), Proposal 1; FCC ECFS, GN Docket 26-281 (26110083691 SpaceX, 26110083692 Amazon) and GN Docket 23-65, filing lists on 2026-10-28
Petition or letter?
Press reports call the carriers' filing a petition. In the FCC's filing system it is an ex parte letter and notice. The FCC's system blocks automated reading of these filings, so this page describes them from their filing records and press reports and quotes none of them.
Source:FCC ECFS, GN Docket 26-281, filings 26110083526 and 26110083511 (CTIA), read through the FCC public API on 2026-10-28
3What the FCC says, and what the vote does
The vote opens comments; it sets no final rules
The Commission votes on a Notice of Proposed Rulemaking, whose issues "remain under consideration and subject to change". It would grow the satellite-to-phone (SCS) bands from five to 13. Comments are due 60 days after Federal Register publication and replies 90 days after; final rules need a later order.
Source:FCC draft NPRM, cover note, fact sheet and filing dates
The FCC already calls TDD the hard case
The draft says TDD "could potentially create coexistence challenges" and asks "whether it is technically possible to achieve such timing synchronization". It notes its proposed limits protect phones, not TDD towers, and that "the current 3GPP standards for NTN focus mainly on FDD, rather than TDD".
This is arithmetic from published inputs, not a new simulation. Each input is cited so you can redo it. It shows the size of the timing problem, not whether interference would be harmful; that needs the simulations listed above.
Across one satellite beam, arrival times can spread over nine NR symbols
A TDD network switches every cell between downlink and uplink at the same instants, with flexible symbols between them as a gap. A satellite can time its signal to arrive on schedule at one point, but not everywhere in its beam, because the distance to the satellite differs across the beam.
Step
Value
Input
One-way delay straight down from 600 km
600 km / 299,792 km/s = 2.00 ms
Altitude used in TR 38.821
Largest delay difference in a 100 km low orbit cell
0.654 ms round trip, so 0.327 ms one way
TR 38.821 V16.2.0, Table 7.2.1.1.1.2-1 (LEO600)
One NR symbol at 30 kHz spacing
0.5 ms slot / 14 symbols = 35.7 µs
TS 38.211 V19.5.0, clause 4.3.2, Table 4.3.2-1
Spread in symbols, worst case
0.327 ms / 35.7 µs = 9.2 symbols
Arithmetic
Same cell with the satellite overhead
Edge 50 km off center: 2.08 km longer path, 6.9 µs, 0.2 symbols
So the spread runs from a fraction of a symbol, with the satellite overhead, to about nine symbols in the worst case TR 38.821 tabulates, which corresponds to a path difference of about 98 km across the cell. Every 10.7 km of path difference costs one symbol. For a 200 km cell the same table gives 1.306 ms round trip, about 18 symbols.
What this does and does not show
It shows that keeping a satellite inside a terrestrial TDD schedule depends on the beam size and elevation angle, and that in the worst case the spread is about nine symbols, two thirds of a 14-symbol slot. It does not show how much interference results: that depends on power, the operators' actual TDD patterns, and how many satellites and phones are active. We have not found a public figure for the gap that US n41 or n77 networks use, so we do not assume one.
Uncertainty: flat ground, one satellite, no timing pre-compensation beyond one reference point. Satellite motion adds a changing delay that we have not included.
3.45 GHz is outside the band range the ITU is studying
The FCC draft describes the WRC-27 studies as covering satellite allocations between 694/698 MHz and 2.7 GHz, downlink only. The 2.5 GHz band ends at 2690 MHz, inside that range. The 3.45 GHz band (3450 to 3550 MHz) is above it, so those studies, as the FCC describes them, do not cover it.
Written with AI assistance from the FCC's draft items, FCC filing records, and the 3GPP specifications and TDocs indexed by 3GPP Scout, and edited by Chris Acker. A second pass checked every number and quote on this page against its source text. Results are the submitting companies' own and were not agreed by 3GPP. Figures from filings we could not read are labeled as claims.
This product uses the FCC Data API but is not endorsed or certified by the FCC.