Confidential client deliverable · Enter the passcode you were given
Imagine Wireless · Powered by OoklaConfidential
PHX · Rev 1 Baseline
Venue Network Intelligence Report · Rev 1 · First issue
Network baseline for Phoenix Sky Harbor across Terminals 3 and 4
The newest passenger building at PHX is the weakest place measured, for every carrier. Over the
footprint of the eight-gate concourse that opened in 2022 (Gates D11 to D18), the healthy tile shares are
T-Mobile 16%, Verizon 9%, AT&T 57%. Away from the buildings the three carriers are close together: the
gate-area 5G medians span 1.9 dB and all sit above the healthy line. The City of Phoenix is building an
airport-wide cellular distributed antenna system through spring 2027, so this window is a baseline.
110
Gate points measured · Concourse A 28 · B 30 · C 16 · D 16 · E 10 · F 10 · of 117 published gates
Gate-carrier pairs below the critical line (−110 dBm)
46.5%
Wi-Fi share of observed MB on tiles in a concourse zone (n=14,851) · 47.1% over the whole clip (n=26,948) · not attributable to any one network
Prepared by the Imagine Wireless Network Intelligence Platform · Powered by Ookla
Data window 2025-09-01 to 2026-08-31 · Issued 2026-09-16 (rev 1) · Freshness 16 days at issue
Prepared for City of Phoenix Aviation Department stakeholder review
Confidential
Section 1
Executive readout
Weakest building · baseline purpose
The newest passenger building at PHX is the weakest place measured, for every carrier. Over the
footprint of the eight-gate concourse that opened in 2022 (Gates D11 to D18), the 5G medians are
T-Mobile −102.0 dBm, Verizon −106.0 dBm, AT&T −94.0 dBm, and the healthy tile shares are T-Mobile 16%,
Verizon 9%, AT&T 57%. Two of the three carriers fail more than four fifths of the tiles there.
Across the two terminal footprints the picture is better and it is not uniform: over the Terminal 3
footprint T-Mobile leads (T-Mobile 92%, Verizon 82%, AT&T 69% healthy), and over the Terminal 4 footprint
Verizon leads (T-Mobile 77%, Verizon 84%, AT&T 74%). Away from the buildings the three carriers are close
together: the gate-area 5G medians span 1.9 dB (T-Mobile −89.7, Verizon −90.2, AT&T −91.5 dBm, n=110 gate
points), all above the −95.0 dBm healthy line, and no gate point is weak on all three carriers. It measures
110 gate points, 117 venue points, 27,206 unique tiles (50,313 carrier-tile observations) and 26,948 usage
tiles over 12 months.
−89.7 dBm
T-Mobile gate-area 5G median · healthy at 81 of 110 gate points (73.6%) · 10 within 1 dB of the line · at risk at 29 · critical at 0
−90.2 dBm
Verizon gate-area 5G median · healthy at 93 of 110 gate points (84.5%) · 6 within 1 dB of the line · at risk at 17 · critical at 0
−91.5 dBm
AT&T gate-area 5G median · healthy at 88 of 110 gate points (80.0%) · 16 within 1 dB of the line · at risk at 22 · critical at 0
What a gate-area number is at PHX
Every gate figure on this page is an aggregate of the tiles within 80 m of a gate point, and at PHX that
circle reaches apron and roadway as much as building. Across the 110 gate points, a median of
21.6% of the tile centroids inside the join fall within a building footprint, and the highest at
any gate is 37% (per concourse: A 20.6%, B 22.9%, C 16.5%, D 20.1%, E 21.6%, F 26.6%). A gate-area figure
is therefore mostly an outdoor ramp reading. Gate-area and zone results in this report are reported as
counts and shares and are never called a lead; the report names a leader only over a building
footprint (section 4.7).
External benchmark, engaged with up front
Ookla's worldwide airport internet performance report for the first quarter of 2025 named PHX one of only
three airports out of 48 that met the Federal Communications Commission benchmark of 100 Mbps download
and 20 Mbps upload on both in-airport Wi-Fi and mobile not independently verified.
That is not in tension with this report. It measures throughput on the public networks; this
report measures signal level and signal quality and carries no throughput or latency field at all
(section 16). A venue can clear a download benchmark and still hold the gaps measured here, because a
strong signal can be congested and a weak signal can be fast when few devices share it. The two readings
answer different questions, and the one this report answers is where the radio conditions are weakest and
which buildings they sit in.
Provenance legend: measured values compute from the Ookla exports at build time. Facts marked "held"
come from a source document pinned in this build and checksummed in the verification bundle. Facts marked
"not independently verified" are public record that this build cites without holding a copy. Facts marked "Imagine knowledge graph"
come from the pinned Imagine knowledge-graph snapshot. Section 16 defines every threshold and basis.
Phoenix Sky Harbor International Airport is operated by the City of Phoenix Aviation Department
held. Two terminals are in service: Terminal 3, with two concourses, the North
Concourse at gates E1 to E10 and the South Concourse at gates F1 to F15, after a 590 million dollar
modernization completed in 2020, all gates common use; and Terminal 4, with eight concourses on Level 3,
four on the north side carrying the A and B gates and four on the south side carrying the C and D gates.
The official Level 3 map, revision October 2, 2022, assigns gates A1 to A30 and B1 to B22 to American
Airlines, gates B23 to B28 to international carriers, and gates C1 to C19 and D1 to D8 to Southwest
Airlines; gates D11 to D18 in the 2022 concourse are Southwest. Airlines that began service after that map,
including China Airlines and STARLUX Airlines, are not assigned to a concourse in any held source. Terminal
2 closed on February 4, 2020 after 58 years of service and its site now holds aircraft parking positions,
so two terminals carried the whole window all held.
T-Mobile
Tied with Verizon at the gate-area 5G median, healthy at the fewest gate points of the three, and the
weakest LTE layer. T-Mobile's venue 5G gate median is −89.7 dBm, 0.5 dB from Verizon's and inside
the 1 dB noise floor, so the two are reported as tied. It holds healthy 5G reference signal received power
(RSRP) at 81 of 110 gates (73.6%; 10 gates within 1 dB of the line), the fewest of the three, with 29 gates
at risk and 0 critical. It holds the strongest gate-area 5G reading at 46 of 110 gates (Concourse A 9 of
28, Concourse B 25 of 30, Concourse C 6 of 16, Concourse D 0 of 16, Concourse E 0 of 10, Concourse F 6 of
10, counts), concentrated in Concourse B, and reports 5G on 92% of its on-airport tiles, a share that mixes
deployment with the panel's handset mix (section 16) and is not compared across carriers here. Its
signal-quality median is +10.2 dB (61 of 110 gates healthy), the lowest of the three. Its lowest zone is
Terminal 4 Concourse D (44% of tiles healthy) and over the Gates D11 to D18 footprint it is healthy on 16%
of tiles. Its LTE layer is the weakest of the three (gate median −93.8 dBm, 60 of 110 gates healthy).
Verizon
Tied with T-Mobile at the gate-area 5G median, healthy at the most gate points, and holding the highest
5G signal-quality median on a C-band-dominant footprint. Verizon's 5G gate median is −90.2 dBm. It
is healthy at 93 of 110 gates (84.5%; 6 gates within 1 dB of the line), the most of the three, with 17
gates at risk and 0 critical, and its 5G signal-quality median of +16.2 dB (103 of 110 gate points healthy)
is the highest of the three, read with the band confound below. It holds the strongest gate-area 5G reading
at 29 of 110 gates and reports 5G on 80% of its tiles. Its LTE layer is healthy (gate median −89.5 dBm, 86
of 110 gates). Its lowest zone is Terminal 4 Concourse D (54% healthy), and over the Gates D11 to D18
footprint it is healthy on 9% of tiles, the lowest reading in the building-masked table. Verizon is the
weakest carrier at 33 of 110 ranked asymmetry gates.
AT&T
Third at the gate-area 5G median by 1.4 dB, holding the strongest LTE reading at 64 of 110 gate points,
and the least weak of the three over the newest Terminal 4 building's footprint, which is the weakest
building measured. AT&T's 5G gate median is −91.5 dBm, 1.4 dB behind Verizon and 1.9 dB behind
T-Mobile, above the healthy line. 88 of 110 gates (80.0%; 16 gates within 1 dB of the line) are healthy, 22
at risk, 0 critical. It reports 5G on 48% of its on-airport tiles, the smallest footprint; the other 52%
carry LTE only. Its LTE layer is the strongest of the three: best-LTE at 64 of 110 gates, gate median
−88.6 dBm, 102 of 110 gates healthy. Its 5G signal-quality median is +13.8 dB (98 of 110 gates healthy).
AT&T holds the strongest gate-area 5G reading by count in Concourse A, Concourse C, Concourse D and
Concourse E (Concourse A 10 of 28, Concourse C 7 of 16, Concourse D 11 of 16, Concourse E 6 of 10, counts),
and over the Gates D11 to D18 footprint it is healthy on 57% of tiles against T-Mobile 16% and Verizon 9%.
Its lowest zone is Terminal 3 Concourse F (South) (57% healthy). AT&T is the weakest carrier at 49 of 110
ranked asymmetry gates.
Per terminal
Terminal 4 gate medians: T-Mobile −89.4, Verizon −90.3, AT&T −91.5 dBm (n=90 gates); healthy 5G at
T-Mobile 61 of 90 gates (68%), Verizon 75 of 90 gates (83%), AT&T 73 of 90 gates (81%). Terminal 3 gate
medians: T-Mobile −90.2, Verizon −89.2, AT&T −91.8 dBm (n=20 gates, below the 30-gate floor, counts only);
healthy 5G at T-Mobile 20 of 20 gates, Verizon 18 of 20 gates, AT&T 15 of 20 gates (counts).
A different carrier leads each building footprint
Over the OpenStreetMap building footprints, which is the only basis on which this report names a leader:
Terminal 3 T-Mobile leads (T-Mobile 92%, Verizon 82%, AT&T 69% of tiles healthy).
Terminal 4 Verizon leads (T-Mobile 77%, Verizon 84%, AT&T 74%). Gates D11 to D18, the 2022
eighth concourse occupied by Southwest Airlines, AT&T leads (T-Mobile 16%, Verizon 9%, AT&T 57%).
Every ranked cell clears the 30-tile floor. The leader changes from building to building, which is the
pattern a phased in-building program produces and equally the pattern macro-site geometry can produce;
section 4.7 carries the full table and section 13 asks which it is.
Photo · Terminal 4 gate hold room, August 2023Wikimedia Commons
Terminal 4 gate hold room, August 2023. Photo: FASTILY, CC BY-SA 4.0, via Wikimedia Commons. Illustrative, resized only.
How the order holds
Gate-median order: T-Mobile and Verizon tied, then AT&T. The T-Mobile-to-Verizon margin is
0.5 dB, under the 1 dB noise floor set in section 16, so this report never ranks one above the
other on that figure; T-Mobile is listed first only because its median is numerically higher. AT&T
trails Verizon by 1.4 dB, at or above the floor, so it is ranked third on the gate median. On the
healthy-gate count the order is Verizon, AT&T, T-Mobile (93, 88, 81 of 110); on signal quality it is
Verizon, AT&T, T-Mobile. All three medians sit above the −95 dBm line. The order reproduces from the
tile export alone.
Why the baseline matters
The City of Phoenix runs an airport-wide cellular distributed antenna system (DAS) program that covers
Terminals 3 and 4, the Rental Car Center, the parking garages, the cell-phone and taxi lots, the PHX Sky
Train stations and adjacent aviation buildings, with AT&T, Verizon and T-Mobile named as the carrier
team held. The City's project page records construction and installation as
ongoing and continuing through spring 2027, so the system was being built across the whole analysis
window held. The City procured it as revenue contract solicitation
AVN RCS 21-008, issued in December 2020 with responses due March 2, 2021, for a third party to design,
install, operate, manage and maintain a turnkey DAS for all wireless service providers
held. The reconciled Airport DAS Master (2026-07-28) records PHX as DAS yes,
self-owned no, neutral host yes, operator a carrier consortium led by AT&T Mobility, contract term 2025
to 2035 at year precision, with the field note that installation of a new system is under way, and a
Phoenix Aviation Advisory Board agenda packet (2026-02-19) describes the Sky Harbor DAS as an AT&T
revenue contract with construction running through spring 2027 across Terminals 3 and 4, the Rental Car
Center, the garages, and the Sky Train stations, funded by the contractor on a revenue-share basis rather
than as a capital line Imagine knowledge graph. The public record names the carrier team
and the campus-wide scope, but it does not say which buildings carry service on the new system in this
window, or which bands each carrier has on it, so nothing in this report attributes any carrier's
reading to the DAS. What the record does establish is that the venue is mid-installation: this window
is the baseline reading, and section 14 writes the per-building criteria the after reading is measured
against. Because which buildings already carried service is not held, the baseline may blend served and
unserved space.
What the pattern is consistent with (inference)
Band first. Verizon's strongest-serving 5G band is NR3700 on 74% of its 9,798 band-reporting 5G tiles;
AT&T's is NR3700 on 82% of 1,366 (a smaller cell-resolved sample, section 9.1); T-Mobile's is NR2500 on
63% of 15,875. Three healthy gate medians over low-rise, wide-apron concourses is the expected macro
propagation picture for this venue: the 80 m gate join reaches apron and roadway as much as building. The
building-masked view is where the carriers separate: over the Gates D11 to D18 footprint AT&T is healthy
on 57% of tiles against 9% for Verizon and 16% for T-Mobile, and over Terminal 3 T-Mobile is healthy on
92% against 82% and 69%. A per-building split of that shape is consistent with in-building sources that
serve one carrier and not the others, which is what a DAS in phased installation would produce, and
equally with macro-site geometry. The export cannot separate the two, and this report does not state
that any carrier is on any in-building system.
How to read the order
Every ranking in this report is signal-layer: RSRP and signal-to-interference-plus-noise ratio (SINR)
from an Android crowdsource panel. The export contains no throughput or latency, so it measures
where signal is. It does not measure whether a call or app fails, and a strong-signal network can
still be congested. The panel's composition differs by carrier and is Android-only, and handset vintage
decides which devices can report C-band 5G (Verizon, AT&T) versus n41 (T-Mobile) at all, so
cross-carrier comparisons inherit a device-population bias the export cannot correct. Nothing in this
report is measured indoors; gate and tenant values are tiles within 80 m of a point (section 16).
Section 2
Venue snapshot and asset inventory
Operator · gate points · in-building record
Provenance legend: measured values compute from the Ookla exports at build time. Facts marked "held"
come from a source document pinned in this build and checksummed in the verification bundle: the City of
Phoenix Aviation Department page, the airport's history, facts, statistics and airline pages, the official
Terminal 4 Level 3 gate map, the Terminal 3 modernization media kit, the City's cellular distributed
antenna system project page, the airport Wi-Fi page, and a Phoenix Aviation Advisory Board agenda packet.
Facts marked "not independently verified" are public record that this build cites without holding a copy. Facts marked
"Imagine knowledge graph" come from the pinned Imagine knowledge-graph snapshot.
Element
Value
Notes
Airport / IATA / ICAO
Phoenix Sky Harbor International Airport / PHX / KPHX
FAA large hub not independently verified; 51,618,649 total passengers in 2025, -1.4% versus 2024 (airport statistics page held; passengers in both directions, a different basis from the FAA enplanement figure used in section 12)
Operating authority
City of Phoenix Aviation Department
A municipal department run as an enterprise fund of the City of Phoenix; major contracts and leases go to City Council on the recommendation of the nine-member Phoenix Aviation Advisory Board, and the airport states that no tax dollars fund its capital program. Terminal 4 and Terminal 3 are both operated directly by the department, with no private terminal lessee held
Terminals analyzed
2
Terminal 3: Advanced Air, Air Canada, Alaska Airlines, Allegiant, Breeze Airways, Delta Air Lines, Denver Air Connection, Frontier Airlines, Hawaiian Airlines, JetBlue Airways, Porter Airlines, Southern Airways Express, Sun Country Airlines, and United Airlines; two concourses, the North Concourse with gates E1 to E10 and the South Concourse with gates F1 to F15, after a 590 million dollar modernization completed in 2020; all gates are common use. Terminal 4: Aeromexico, Air France, American Airlines, British Airways, China Airlines, Contour Airlines, Southwest Airlines, STARLUX Airlines, Volaris, and WestJet; eight concourses on Level 3, four on the north side carrying the A and B gates and four on the south side carrying the C and D gates. The official Level 3 map, revision October 2, 2022, assigns gates A1 to A30 and B1 to B22 to American Airlines, gates B23 to B28 to international carriers, and gates C1 to C19 and D1 to D8 to Southwest Airlines; gates D11 to D18 in the 2022 concourse are Southwest. Airlines that began service after that map, including China Airlines and STARLUX Airlines, are not assigned to a concourse in any held source. Terminal 2 closed on February 4, 2020 after 58 years of service and its site now holds aircraft parking positions, so two terminals carried the whole window all held
Zones for tile assignment
6
Concourse A (Terminal 4) · Concourse B (Terminal 4) · Concourse C (Terminal 4) · Concourse D (Terminal 4) · Concourse E (Terminal 3) · Concourse F (Terminal 3) (nearest gate or tenant point within 500 m)
Gate points analyzed
110
Concourse A 28, Concourse B 30, Concourse C 16, Concourse D 16, Concourse E 10, Concourse F 10 (OpenStreetMap inventory pinned 2026-09-16, zoned by the concourse letter in the gate ref and checked against the terminal footprint). The official Terminal 4 Level 3 map, revision October 2, 2022, publishes gates A1 to A30, B1 to B28 with B23 to B28 international, C1 to C9 and C11 to C19, D1 to D8, and the airport's history page adds gates D11 to D18 in the 2022 concourse; Terminal 3 publishes gates E1 to E10 in the North Concourse and F1 to F15 in the South Concourse, with the terminal total stated as 25 gates in the 2020 modernization record and as 20 in a 2025 City release that adds six more held. The measured inventory is smaller than the published one and the reconciliation is tabulated in section 2.1; 1 further OpenStreetMap node on the airport carries the tag but no concourse letter and number (ref 'D', OpenStreetMap id 3081169829) and is excluded from every figure in this report
Mapped venue points
117
50 food · 33 retail · 21 transport_admin · 7 services · 3 lounge · 3 other. Zone attribution: A 23, B 27, C 14, D 8, E 12, F 16, outside zones 17. OpenStreetMap capture, pinned 2026-09-16 transcribed points only. 132 points were captured and 15 were excluded before analysis because they are not venue points: named views of distant landmarks, public art, building fabric, airfield and tower infrastructure, and a cargo operator facility. 6 points were refiled where the capture's category contradicted the point's function, including three airline and card-issuer lounges captured as other. Every call is listed with its reason in the audit dossier
City airport-wide cellular DAS program, under construction through spring 2027; carrier team AT&T, Verizon and T-Mobile
The City of Phoenix runs an airport-wide cellular distributed antenna system (DAS) program that covers Terminals 3 and 4, the Rental Car Center, the parking garages, the cell-phone and taxi lots, the PHX Sky Train stations and adjacent aviation buildings, with AT&T, Verizon and T-Mobile named as the carrier team; the City's project page records construction and installation as ongoing and continuing through spring 2027, so the system was being built across the whole analysis window; the City procured it as revenue contract solicitation AVN RCS 21-008, issued in December 2020 with responses due March 2, 2021, for a third party to design, install, operate, manage and maintain a turnkey DAS for all wireless service providers; the solicitation required that legacy in-building systems, repeaters, bi-directional amplifiers and standalone carrier systems inside the DAS coverage area be decommissioned once the new system carries service all held. The solicitation's coverage exhibit names Terminal 3 levels 1 to 4, the North and South concourses, the curbside and the ramp areas outside each gate, and the same exhibit names Terminal 4 levels 1 to 4 including the boarding bridges, the seven concourses that existed then, and the concourse that was under construction and opened in June 2022 Imagine knowledge graph. The reconciled Airport DAS Master (2026-07-28) records PHX as DAS yes, self-owned no, neutral host yes, operator a carrier consortium led by AT&T Mobility, contract term 2025 to 2035 at year precision, with the field note that installation of a new system is under way; a Phoenix Aviation Advisory Board agenda packet (2026-02-19) describes the Sky Harbor DAS as an AT&T revenue contract with construction running through spring 2027 across Terminals 3 and 4, the Rental Car Center, the garages, and the Sky Train stations, funded by the contractor on a revenue-share basis rather than as a capital line Imagine knowledge graph. Which buildings carry service on the new system in this window, and which bands each carrier has on it, are open items (section 8)
Knowledge-graph coverage
Profile, DAS and Boingo deployments, RFP and board evidence
Profile: large hub (the graph also holds an enplanement count with no year recorded; it is not published here, because the dated FAA calendar-2024 figure in section 12 is the report's single enplanement basis). Deployments: AT&T Mobility DAS; Boingo Wireless Wi-Fi; Boingo Wireless Passpoint. Contracts query not available; no leadership rows Imagine knowledge graph
Public Wi-Fi
SSID Free PHX Boingo WiFi
Boingo Wireless under City of Phoenix contract 152010, a revenue contract covering the public areas of both terminals and the Rental Car Center; the City extended it through July 31, 2026 and then month to month while it transitions the service in house held; the City reports more than 37,000 users a day on the public Wi-Fi and more than 13.7 million users across 2024 held; the graph also holds active Boingo Wireless Wi-Fi and Passpoint deployments Imagine knowledge graph
Capital program
A new west terminal between Terminal 3 and Taxiway U is in planning, the first new terminal since 1990, with construction indicated for 2029 to 2033 and an opening in 2033 at a figure the City has put at around 4 billion dollars; an 84.3 million dollar federal grant announced in January 2025 funds the new Taxiway U bridge
The 310 million dollar Terminal 4 eighth concourse, eight gates for Southwest Airlines, opened to the public on June 20, 2022; the 590 million dollar Terminal 3 modernization finished in 2020, and a further concourse on the north side of Terminal 3 is under construction for six more gates at about 368 million dollars including enabling work, targeted to open in 2027; an expanded lower-level gate area in the Terminal 3 South Concourse opened on July 22, 2026, inside this window; stage 2 of the PHX Sky Train, a 745 million dollar extension to the Rental Car Center with new 24th Street and Rental Car Center stations, opened on December 20, 2022 and removed 84 rental-car shuttle buses from the roadways not independently verified
Ground transport
The PHX Sky Train, a free 24-hour automated people mover with six stations at 44th Street and Washington, East Economy, Terminal 4, Terminal 3, 24th Street and the Rental Car Center
Context only not independently verified
Recognition
Third among mega airports in the J.D. Power North America Airport Satisfaction Study in both 2025 and 2026, and top-ten placings for shopping, dining and large airports in the 2025 USA Today 10Best readers' choice awards
Context only not independently verified
Photo · Aerial view of Sky Harbor, May 2025Wikimedia Commons
Aerial view of Sky Harbor, May 2025: Terminal 4 with its concourses in the foreground, Terminal 3 beyond. Illustrative only. Photo: Designism, CC0, via Wikimedia Commons. Resized only.
Venue map: aerodrome outline, OSM terminal buildings, 110 gates with concourse letters, and the 50,313 on-airport RF tile centroids. North up; scale bar in meters. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Photo · PHX Sky Train at the Terminal 4 station, October 2023Wikimedia Commons
PHX Sky Train at the Terminal 4 station, October 2023. The Sky Train stations are named in the DAS coverage exhibit. Photo: Hunter Trick (TrickHunter), CC BY-SA 4.0, via Wikimedia Commons. Illustrative, resized only.
2.1 Measured gate inventory against the published gate list
The analysis uses the 110 measured points. Of the 117 gates on the published lists, 10 have no measured
point in the pinned OpenStreetMap inventory, so a per-concourse count in this report is a count of
measured points and is written that way throughout. Points carried in OpenStreetMap that are not on the
published list (the B1A, B15A and B15B hardstands) are kept and labelled with their OpenStreetMap ref.
Aligning the two lists is a request to the Aviation Department (section 15), and no gate-named criterion in
section 14 should be adopted before that is done.
Every cellular number in this report is an aggregate of crowdsourced background scans from consumer
Android devices (about 3% of handsets; iOS devices are not sampled), binned by Ookla to map tiles of about
9 m by 11 m, then joined to gate and tenant points. The numbers describe where devices were over a
12-month window. They are not drive-test measurements. Every n counts tiles, and every table carries its
n so thin cells are visible.
3.0 Carrier summary
Carrier
5G RSRP median
5G SINR median
LTE RSRP median
Healthy 5G gates
n (gates / RF tiles)
T-Mobile
−89.7 dBm
+10.2 dB
−93.8 dBm
81 of 110
110 / 18,918
Verizon
−90.2 dBm
+16.2 dB
−89.5 dBm
93 of 110
110 / 18,513
AT&T
−91.5 dBm
+13.8 dB
−88.6 dBm
88 of 110
110 / 12,882
Healthy = gate mean ≥ −95 dBm. SINR healthy (≥ 10 dB): Verizon 103, AT&T 98, T-Mobile 61 of 110 gates.
LTE healthy gates: AT&T 102, Verizon 86, T-Mobile 60 of 110. Strongest LTE reading: AT&T 64, Verizon 36, T-Mobile 10 gates. Full per-gate tables ship in the PDF appendices.
5G footprint and conditional medians (tile basis). Each 5G median is conditional on the tile reporting
a 5G reading; the last column combines presence and level.
Venue medians across 110 gates against the locked threshold ladder (section 16). The two received-power panels draw bars from −115 dBm, so a longer bar is a stronger signal. Locked ladder: healthy ≥ −95 · at-risk −95 to −110 · critical < −110 dBm. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Fig. 3 · SINR distribution per carrier at gate-point levelSource · Ookla
SINR distribution per carrier at gate-point level. Read across carriers with care: SINR depends on the serving band's bandwidth and on neighbour loading, and these carriers sit on different bands here (section 9.1 gives each carrier's strongest-serving band mix), so a cross-carrier SINR gap is not a like-for-like comparison. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Cross-carrier SINR comparison, stated once and applied throughout
The three carriers are not on the same spectrum at this venue. Section 9.1 shows the strongest-serving
band mix, and a carrier whose 5G sits mostly on a wide mid-band channel will not produce an SINR
distribution comparable to one sitting mostly on a different band with different neighbour loading and a
different receiver population. Within a carrier, SINR across zones is a fair comparison and is used
that way; across carriers it is context, not a ranking.
Fig. 4 · 5G NR RSRP by tile and carrier · terminal area · locked coverage classSource · Ookla
Terminal-area map of 5G NR RSRP by tile and carrier, colored by the locked coverage class. Grey tiles carry no 5G reading. Each tile is about 9 m by 11 m; n is the tile count per panel. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
What each carrier's readings show
What the T-Mobile readings show
Tied with Verizon at the gate-area 5G median, healthy at the fewest gate points of the three (81 of 110),
and the lowest 5G SINR median (+10.2 dB, on a different band mix from the other two). Its strongest place
is over the Terminal 3 footprint (92% of tiles healthy) and its weakest is over the Gates D11 to D18
footprint (16%, median −102.0 dBm). On the zone basis its lowest concourse is Concourse D. A
re-measurement over the Gates D11 to D18 footprint is the clearest test of whether the new system carries
T-Mobile there.
What the Verizon readings show
Tied with T-Mobile at the gate-area 5G median, healthy at the most gate points (93 of 110), and the
highest 5G SINR median of the three (+16.2 dB, on a different band mix from the other two, so that gap is
context and not a ranking). Its one weak place is over the Gates D11 to D18 footprint, where it is
healthy on 9% of tiles with a median of −106.0 dBm, the lowest cell in the building-masked table.
That footprint is therefore where a re-measurement would show the largest change for Verizon if the
new system carries service there.
What the AT&T readings show
Third at the gate-area 5G median by 1.4 dB, and the strongest LTE layer of the three (64 of 110 gate
points, median −88.6 dBm). Its 5G reports on 48% of its tiles, the smallest share of the three, which
this report does not read as a coverage figure because the share is confounded by the panel's handset mix
(section 16). Over the Gates D11 to D18 footprint it is the least weak of the three (57% of tiles
healthy, median −94.0 dBm), in a building where the other two are at 16% and 9%; over the Terminal 3 and
Terminal 4 footprints it is the weakest of the three (69% and 74%). The knowledge graph names AT&T
Mobility as the lead of the DAS carrier consortium Imagine knowledge graph;
whether that explains the pattern is a Recommendation 1 question, not a finding here.
3.1 T-Mobile results
T-Mobile is tied with Verizon at the top of the 5G gate median (−89.7 dBm, 0.5 dB apart). 81 of 110
gates (73.6%; 10 gates within 1 dB of the line) are healthy, 29 at risk, 0 critical; on that count it
trails Verizon (93) and AT&T (88). It is the strongest 5G reading at 46 of 110 gates, 25 of them in
Concourse B (T-Mobile 25, Verizon 4, AT&T 1; n=30 gates). SINR is +10.2 dB at the median (61 of 110 gates
healthy), the lowest of the three. It reports 5G on 92% of its on-airport tiles (17,312 of 18,918); Verizon
reports 5G on 80% and AT&T on 48%. Its LTE layer is the weakest of the three: LTE gate median −93.8 dBm,
healthy at 60 of 110 gates, strongest LTE reading at 10 gates.
Healthy tile shares by zone: Concourse A 60%, Concourse B 92%, Concourse C 77%, Concourse D 44%,
Concourse E 67%, Concourse F 58% (n ≥ 994 tiles per zone). Its lowest zone is Terminal 4 Concourse D. Over
the pinned building footprints (section 4.7) it is healthy on 92% of Terminal 3 tiles, 77% of Terminal 4
tiles, and 16% of the Gates D11 to D18 building. Fig. 4 shows the pattern: green along Concourse B and
Terminal 3, amber over the east end of Terminal 4. Its terminal-vs-periphery proxy is −1.0 dB
(section 5.2). T-Mobile is the weakest carrier at 28 of 110 ranked asymmetry gates.
3.2 Verizon results
Verizon's 5G gate median is −90.2 dBm, tied with T-Mobile (0.5 dB apart) and 1.4 dB ahead of AT&T.
93 of 110 gates (84.5%; 6 gates within 1 dB of the line) are healthy, the most of the three, with 17 at
risk and 0 critical. Its 5G signal quality is the best of the three: SINR median +16.2 dB, healthy at 103
of 110 gates. It is the strongest 5G reading at 29 of 110 gates and reports 5G on 80% of its on-airport
tiles (14,734 of 18,513). Its LTE layer is healthy: LTE gate median −89.5 dBm (86 of 110 gates), strongest
LTE reading at 36 gates.
Healthy tile shares by zone: Concourse A 63%, Concourse B 78%, Concourse C 66%, Concourse D 54%,
Concourse E 72%, Concourse F 75% (n ≥ 915 tiles per zone). Its lowest zone is Terminal 4 Concourse D. Over
the footprints Verizon is healthy on 84% of Terminal 4 tiles and 82% of Terminal 3 tiles, but on 9% of the
Gates D11 to D18 building (median −106.0 dBm, n=98), the lowest cell in the building-masked table
(section 4.7). Verizon's 5G is C-band (NR3700 on 74% of its band-reporting 5G tiles), so a healthy gate
median with one weak building is consistent with macro C-band that reaches the apron-side gates and not the
interior of the newest building; whether that building has a live in-building source for any carrier is a
Recommendation 1 question. Its terminal-vs-periphery proxy is +2.0 dB (section 5.2). Verizon is the weakest
carrier at 33 of 110 ranked asymmetry gates.
3.3 AT&T results
AT&T's 5G gate median is −91.5 dBm, 1.4 dB behind Verizon and 1.9 dB behind T-Mobile, and above the
healthy line. 88 of 110 gates (80.0%; 16 gates within 1 dB of the line) are healthy, 22 at risk, 0
critical. SINR is +13.8 dB at the median (98 of 110 gates healthy). It is the strongest 5G reading at 35
of 110 gates, leading the count in Concourse A, Concourse C, Concourse D and Concourse E (Concourse A
AT&T 10, T-Mobile 9, Verizon 9, n=28; Concourse C AT&T 7, T-Mobile 6, Verizon 3, n=16; Concourse D AT&T 11,
Verizon 5, n=16; Concourse E AT&T 6, Verizon 4, n=10; counts). AT&T is the weakest carrier at 49 of 110
ranked asymmetry gates.
The 5G footprint is the smallest of the three: AT&T reports 5G on 48% of its on-airport tiles
(6,179 of 12,882); tiles with an LTE reading but no 5G reading are 49% of its tiles. AT&T's LTE layer
carries its own service well here: strongest LTE reading at 64 of 110 gates, LTE gate median −88.6 dBm,
healthy at 102 of 110 gates. Healthy tile shares by zone: Concourse A 58%, Concourse B 58%, Concourse C
71%, Concourse D 67%, Concourse E 73%, Concourse F 57% (n ≥ 440 tiles per zone); its lowest zone is
Terminal 3 Concourse F (South). Over the footprints AT&T is healthy on 57% of the Gates D11 to D18 building
(median −94.0 dBm, n=68), the highest of the three there, against 74% of Terminal 4 and 69% of Terminal 3,
the lowest of the three in both. AT&T's 5G is NR3700 on 82% of 1,366 band-reporting 5G tiles (a smaller
cell-resolved sample). On Fig. 4, AT&T's panel shows green over the east end of Terminal 4 and amber over
Terminal 3 and Concourse B; the building-masked view (section 4.7) quantifies it per building.
3.4 5G footprint and coverage-band attachment
Fig. 5 · 5G footprint · share of tiles with a 5G NR readingSource · Ookla
Share of each carrier's on-airport tiles that report a 5G NR reading (5G footprint). CONFOUNDED: a tile reports 5G only if a handset in the panel could report that carrier's 5G band, so this share mixes the carrier's 5G deployment with the mix of 5G-capable handsets that carrier has in an Android crowdsource panel. It is not a coverage measurement and the carriers are not comparable on it (section 16). Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
LTE percent-low-band per carrier: mean, share of tiles above zero, and p90. Higher means more attachment to sub-1 GHz coverage bands. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
The percent-low-band statistic describes LTE coverage-band attachment. A tile that attaches to sub-1 GHz
spectrum is using a coverage-grade band, which propagation, carrier idle-mode priorities, and site geometry
all drive. T-Mobile mean 1.7% (tiles above zero: 6% of n=18,763; p90 0%) · Verizon mean 12.7% (tiles above
zero: 36% of n=18,271; p90 50%) · AT&T mean 14.2% (tiles above zero: 39% of n=12,558; p90 50%).
The LTE metric does not see 5G low band. On the 5G side: T-Mobile: NR600 on 666 tiles, 4.2% of its
15,875 band-reporting NR tiles. Verizon: NR600 on 1 tile, 0.0% of its 9,798 band-reporting NR tiles, and
NR850 on 723 tiles, 7.4% of the same 9,798. AT&T: NR600 on 1 tile, 0.1% of its 1,366 band-reporting NR
tiles, and NR850 on 12 tiles, 0.9% of the same 1,366. Section 9.1 lists the bands.
3.5 RSRQ, channel quality (CQI), and optimization priority
RSRQ medians: T-Mobile NR −11.0 / LTE −13.0 dB; Verizon NR −11.0 / LTE −12.0 dB; AT&T NR −11.0 /
LTE −13.0 dB. Every carrier's medians sit below the -10 dB line on both technologies. Medians of −11 to
−14 dB are ordinary loaded-network values, and the line is a generic reference, not a venue-specified
threshold, so this is a venue-wide observation to read together with SINR, not a failure call. RSRQ
degrades as subframe utilization rises, so a weak RSRQ on a strong-RSRP network is consistent with a
loaded serving layer. This is an inference; the export carries no load field (section 16).
CQI medians: T-Mobile median 7.0 (n=4,222); Verizon median 9.0 (n=3,875); AT&T median 9.0 (n=4,726).
CQI is the closest throughput proxy in an export that carries no throughput fields. Ookla's LTE
optimization priority is the vendor's own index for tiles where signal is strong but quality is poor. Venue
medians: T-Mobile median 62 (n=8,310); Verizon median 60 (n=13,655); AT&T median 61 (n=10,462). The
discriminating signal is in the distribution tails. The RSRQ, CQI, and optimization-priority distributions
(Figs. 7 and 8) ship in the PDF.
Where the strongest cell at a tile is not the most-frequent serving cell, the network is handing over
between competing cells. Measured churn: T-Mobile LTE 36% (n=10,872) / 5G 39% (n=15,874); Verizon LTE 35%
(n=15,156) / 5G 23% (n=9,793); AT&T LTE 40% (n=12,214) / 5G 7% (n=1,365). Cell, PCI, and band fields
populate only where Ookla resolved a serving cell, which is why AT&T's 5G n is 1,365 here against 6,179
for RSRP (field population table, section 16). AT&T's 5G churn therefore rests on 1,365 cell-resolved tiles
and is not compared with the other two; its bar is hatched. Churn blends macro-sector overlap with any
in-building system's handoff to the macro layer. This is a proxy; the export carries no handover trace.
Fig. 9 ships in the PDF.
Section 4
The view by concourse zone
Six concourses · tile basis
Concourse zone gate counts (Concourse A 28, Concourse B 30, Concourse C 16, Concourse D 16, Concourse E
10, Concourse F 10): only Concourse B clears the 30-point floor on the gate basis, so its gate shares are
shown as percentages; every other concourse's gate tallies appear as counts and are never ranked. For one
consistent basis, every zone comparison in this section runs on the tile basis: a tile belongs to the zone
of its nearest gate or tenant point within 500 m, which gives hundreds of tiles per carrier per zone. Zones
are the six concourses (A to D at Terminal 4, E and F at Terminal 3), from the concourse letter in each
gate ref.
4.1 Healthy share and median RSRP by zone
These zone figures are not a leader board
A concourse zone is the set of tiles nearest that concourse's gate and venue points within 500 m, so it
holds apron, taxiway and roadway as well as building, and section 4.7 carries the only basis on which
this report names a leader. Read the zone rows as the spread within a zone: the widest
carrier-to-carrier gap in healthy tile share is Concourse B at 34 points, and the narrowest is
Concourse A at 5 points. Lowest zone per carrier on the healthy tile share: T-Mobile in Concourse D
(44%), Verizon in Concourse D (54%), AT&T in Concourse F (57%). The lowest zone differs by carrier.
Zone
T-Mobile healthy tiles
Verizon healthy tiles
AT&T healthy tiles
Median 5G RSRP (dBm) T-Mobile / Verizon / AT&T
Concourse A (T4)
60%n=1,957
63%n=1,865
58%n=927
−93.0 / −92.0 / −93.0
Concourse B (T4)
92%n=2,492
78%n=1,818
58%n=886
−83.0 / −89.0 / −94.0
Concourse C (T4)
77%n=1,258
66%n=1,172
71%n=480
−89.0 / −92.0 / −90.0
Concourse D (T4)
44%n=994
54%n=915
67%n=440
−97.0 / −94.0 / −91.0
Concourse E (T3)
67%n=1,267
72%n=1,052
73%n=505
−92.0 / −92.0 / −86.0
Concourse F (T3)
58%n=2,064
75%n=2,051
57%n=767
−94.0 / −89.0 / −94.0
Healthy share and median on the tile basis; n = tiles per carrier per zone. Healthy = tile 5G RSRP ≥ −95 dBm. Each zone pool includes apron, taxiway and roadway tiles.
Fig. 10 · Healthy 5G tile share by zone and carrierSource · Ookla
Share of zone tiles with healthy 5G RSRP by carrier (tile basis; minimum n per carrier on the axis). Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Fig. 11 · Median 5G RSRP by zone and carrierSource · Ookla
Per-zone median 5G RSRP per carrier (tile basis). Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Fig. 12 · LTE RSRP by tile and carrier · terminal area · locked coverage classSource · Ookla
Terminal-area map of LTE RSRP by tile and carrier, colored by the locked coverage class. Grey tiles carry no LTE reading. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
4.2 Coverage class and PCI diversity
Critical-band gates (below −110 dBm) by carrier: T-Mobile 0, Verizon 0, AT&T 0 of 110 gates (counts).
On the tile basis the critical share peaks at T-Mobile 1% in Concourse D (n=994); Verizon 3% in Concourse D
(n=915); AT&T 4% in Concourse A (n=927). Lowest healthy share in any zone: T-Mobile 44% (Concourse D);
Verizon 54% (Concourse D); AT&T 57% (Concourse F).
Fig. 13 · Coverage-class distribution by concourse zone and carrierSource · Ookla
Coverage-class distribution by concourse zone and carrier (tile basis; T = T-Mobile, V = Verizon, A = AT&T). Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
The PCI picture in each zone reflects which sectors reach it, macro or in-building. Higher diversity with
healthy SINR means several sectors serve the zone cleanly. Higher diversity with poor SINR means they
compete. The raw unique counts are not comparable across carriers, because the cell-resolved samples behind
them differ by more than an order of magnitude (T-Mobile 15,874, Verizon 9,793, AT&T 1,365 tiles), which is
why Fig. 14 draws the bars per 1,000 cell-resolved tiles. Even normalized, this metric is a sampling
statistic as much as a network property: PCI counts accumulate over 12 months across every 5G layer a
carrier runs, so reassignment and multi-carrier layers inflate them. Read it as a prompt to look at SINR
in the same zone, not as a ranking. Fig. 14 ships in the PDF.
4.3 Cross-carrier concordance by gate
This split shapes the infrastructure question. A gate weak for all three carriers points to a gate-area
problem for that gate's 80 m circle, which can include apron. A gate weak for exactly one carrier points
to that carrier's path.
62
Gates healthy for all three carriers · of 110
28
Gates weak for exactly one carrier · T-Mobile 11 · Verizon 3 · AT&T 14
20
Gates weak for two carriers · T-Mobile and Verizon 12 · T-Mobile and AT&T 6 · Verizon and AT&T 2
0
Gates weak for all three carriers
At 62 of 110 gates every carrier is healthy, and where a gate is weak it is most often one carrier's path
(28 one-weak gates against 20 with two or more weak). That is a per-carrier picture, not a venue-wide
gap, and it is what the per-building split in section 4.7 also shows: the carrier that is weak changes
from building to building. The concordance does not by itself say which buildings carry service on the new
system; Recommendation 1 asks for the per-building phase map.
Fig. 15 · Carriers below the healthy line per gate, by concourse zoneSource · Ookla
Per concourse zone: how many of the three carriers are below the healthy line at each gate (gate counts). Healthy = gate mean 5G RSRP ≥ −95 dBm. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
4.4 Terminal 4 Concourses A and B
Terminal 4 has eight concourses on Level 3, four on the north side carrying the A and B gates and four on
the south side carrying the C and D gates. The official Level 3 map, revision October 2, 2022, assigns
gates A1 to A30 and B1 to B22 to American Airlines, gates B23 to B28 to international carriers, and gates
C1 to C19 and D1 to D8 to Southwest Airlines; gates D11 to D18 in the 2022 concourse are Southwest.
Airlines that began service after that map, including China Airlines and STARLUX Airlines, are not assigned
to a concourse in any held source held.
Concourse A hosts American Airlines held (gates A1 to A14 and A15 to A30 per
the official Terminal 4 Level 3 map rev. 2022-10-02). It holds 28 of the 110 measured gates. Best-5G
counts: AT&T 10, T-Mobile 9, Verizon 9. Healthy 5G gates: T-Mobile 15 of 28, Verizon 21 of 28, AT&T 18 of
28 (counts). Gate-basis 5G medians: T-Mobile −93.5, Verizon −91.7, AT&T −93.1 dBm. Tile-basis healthy
shares: T-Mobile 60%, Verizon 63%, AT&T 58% (n ≥ 927 tiles per carrier), a carrier-to-carrier spread of 5
points on a pool that includes apron and roadway.
Concourse B hosts American Airlines, with B23 to B28 international held
(gates B1 to B14 and B15 to B22 American Airlines; B23 to B28 international, listing American, British
Airways, Condor, Volaris and WestJet on the 2022 map, with China Airlines and STARLUX joining in-window
and the gate area not stated). It holds 30 of the 110 measured gates. Best-5G counts: T-Mobile 25,
Verizon 4, AT&T 1. Healthy 5G gates: T-Mobile 30 of 30 (100%), Verizon 30 of 30 (100%), AT&T 23 of 30
(77%). Gate-basis 5G medians: T-Mobile −85.7, Verizon −88.3, AT&T −91.5 dBm. Tile-basis healthy shares:
T-Mobile 92%, Verizon 78%, AT&T 58% (n ≥ 886 tiles per carrier), a carrier-to-carrier spread of 34 points
on a pool that includes apron and roadway.
Fig. 16 · Per-gate 5G RSRP · 58 Concourse A and B gatesSource · Ookla
Per-gate 5G RSRP across the 58 Concourse A and B gates (OpenStreetMap inventory). Each bar is a mean over the tiles within 80 m of the gate. Bars rise from −115 dBm, so a longer bar is a stronger signal. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Photo · Southwest Airlines 737 on the Terminal 4 apron, July 2025Wikimedia Commons
Southwest Airlines 737 on the Terminal 4 apron, July 2025. Photo: BeeboMan, CC BY 4.0, via Wikimedia Commons. Illustrative, resized only.
4.5 Terminal 4 Concourses C and D
Concourse C hosts Southwest Airlines held (gates C1 to C9 and C11 to C19 per
the official map). It holds 16 of the 110 measured gates. Best-5G counts: AT&T 7, T-Mobile 6, Verizon 3.
Healthy 5G gates: T-Mobile 16 of 16, Verizon 16 of 16, AT&T 16 of 16 (counts). Gate-basis 5G medians:
T-Mobile −89.5, Verizon −89.8, AT&T −88.3 dBm. Tile-basis healthy shares: T-Mobile 77%, Verizon 66%,
AT&T 71% (n ≥ 480 tiles per carrier), a carrier-to-carrier spread of 11 points on a pool that includes
apron and roadway.
Concourse D hosts Southwest Airlines held (gates D1 to D8 per the official
map, plus D11 to D18, the eight-gate eighth concourse opened 2022-06-20 and occupied by Southwest). It
holds 16 of the 110 measured gates. Best-5G counts: AT&T 11, Verizon 5. Healthy 5G gates: T-Mobile 0 of 16,
Verizon 8 of 16, AT&T 16 of 16 (counts). Gate-basis 5G medians: T-Mobile −97.3, Verizon −96.0,
AT&T −93.0 dBm. Tile-basis healthy shares: T-Mobile 44%, Verizon 54%, AT&T 67% (n ≥ 440 tiles per carrier),
a carrier-to-carrier spread of 23 points on a pool that includes apron and roadway.
Gates D11 to D18, the weakest building measured
Gates D11 to D18 are the 310 million dollar eighth concourse, which opened to the public on June 20,
2022 and is occupied by Southwest Airlines; it is the newest passenger building on the campus and it
also added a post-security connector bridge between the D and A concourses
held. Over that building's footprint the carriers separate most
(section 4.7): T-Mobile 16% healthy, median −102.0 dBm (n=99); Verizon 9% healthy, median
−106.0 dBm (n=98); AT&T 57% healthy, median −94.0 dBm (n=68). Verizon's 9% is the lowest cell in the
building-masked table, and two of the three carriers fail more than four fifths of the tiles there.
Fig. 17 · Per-gate 5G RSRP · 32 Concourse C and D gatesSource · Ookla
Per-gate 5G RSRP across the 32 Concourse C and D gates (OpenStreetMap inventory). Each bar is a mean over the tiles within 80 m of the gate. Bars rise from −115 dBm, so a longer bar is a stronger signal. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Photo · American Airlines 737 at gate B22, May 2025Wikimedia Commons
American Airlines 737 at Terminal 4 gate B22, May 2025. Photo: Designism, CC0, via Wikimedia Commons. Illustrative, resized only.
Photo · Terminal 4 concourse dining area, August 2023Wikimedia Commons
Terminal 4 concourse dining area, August 2023. Photo: FASTILY, CC BY-SA 4.0, via Wikimedia Commons. Illustrative, resized only.
4.6 Terminal 3 Concourses E and F
Terminal 3 has two concourses, the North Concourse with gates E1 to E10 and the South Concourse with
gates F1 to F15, after a 590 million dollar modernization completed in 2020; all gates are common use
held. Concourse E (North) hosts Air Canada, Alaska, Hawaiian and United
primarily held (gates E1 to E10, renovated and opened February 2020; the 2020
media kit names Air Canada, Alaska, Contour, Hawaiian and United as primary users; all Terminal 3 gates are
common-use and shareable across concourses). It holds 10 of the 110 measured gates. Best-5G counts:
AT&T 6, Verizon 4. Healthy 5G gates: T-Mobile 10 of 10, Verizon 10 of 10, AT&T 10 of 10 (counts).
Gate-basis 5G medians: T-Mobile −89.0, Verizon −89.2, AT&T −86.7 dBm. Tile-basis healthy shares:
T-Mobile 67%, Verizon 72%, AT&T 73% (n ≥ 505 tiles per carrier), a carrier-to-carrier spread of 6 points
on a pool that includes apron and roadway.
Concourse F (South) hosts Delta and the other Terminal 3 carriers on common-use gates
held (gates F1 to F15, opened January 2019; the Delta Sky Club is in the South
Concourse; the remaining Terminal 3 carriers, including Delta, Frontier, JetBlue, Allegiant, Breeze, Sun
Country and Spirit until 2026-01-08, use these common-use gates; a per-airline gate assignment is not
held). It holds 10 of the 110 measured gates. Best-5G counts: T-Mobile 6, Verizon 4. Healthy 5G gates:
T-Mobile 10 of 10, Verizon 8 of 10, AT&T 5 of 10 (counts). Gate-basis 5G medians: T-Mobile −91.2,
Verizon −92.2, AT&T −95.2 dBm. Tile-basis healthy shares: T-Mobile 58%, Verizon 75%, AT&T 57%
(n ≥ 767 tiles per carrier), a carrier-to-carrier spread of 18 points on a pool that includes apron and
roadway. Over the Terminal 3 footprint (section 4.7): T-Mobile 92% healthy, median −86.0 dBm (n=344);
Verizon 82% healthy, median −90.0 dBm (n=333); AT&T 69% healthy, median −90.0 dBm (n=227); T-Mobile
leads.
Fig. 18 · Per-gate 5G RSRP · 20 Concourse E and F gatesSource · Ookla
Per-gate 5G RSRP across the 20 Concourse E and F gates (OpenStreetMap inventory). Each bar is a mean over the tiles within 80 m of the gate. Bars rise from −115 dBm, so a longer bar is a stronger signal. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Photo · Alaska Air Group aircraft at Terminal 3, April 2025Wikimedia Commons
Alaska Air Group aircraft at Terminal 3, April 2025. Photo: Ziongarage, CC0, via Wikimedia Commons. Illustrative, resized only.
4.7 Building-masked view (OSM polygons)
The building-masked basis is the only basis on which this report names a terminal or concourse leader,
because the 80 m gate join at PHX reaches apron and roadway as much as building.
Fig. 19 · Building-masked view · 5G NR class for tiles inside the OSM footprintsSource · Ookla
Tiles whose centroid falls inside a pinned OSM building polygon (Terminal 3, Terminal 4, and the separately mapped Gates D11 to D18 building), colored by 5G NR class, with the per-building per-carrier table. This is our OSM building-polygon mask, not Ookla's Indoor layer. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Building (OSM polygon)
T-Mobile · n · healthy · median
Verizon · n · healthy · median
AT&T · n · healthy · median
Leader (tie-aware)
T3: Terminal 3
344 · 92% · −86.0 dBm
333 · 82% · −90.0 dBm
227 · 69% · −90.0 dBm
T-Mobile leads
T4: Terminal 4
546 · 77% · −90.0 dBm
540 · 84% · −89.0 dBm
412 · 74% · −91.0 dBm
Verizon leads
T4: Gates D11-D18
99 · 16% · −102.0 dBm
98 · 9% · −106.0 dBm
68 · 57% · −94.0 dBm
AT&T leads
Healthy = tile 5G RSRP ≥ −95 dBm; median over tiles inside the polygon. n counts tiles, not scans. Every ranked cell clears the 30-tile floor.
The mask is a point-in-polygon test of tile centroids against the pinned OpenStreetMap footprints, so it
selects tiles over the buildings rather than tiles measured inside the buildings (section 16). It is the
basis on which this report names a terminal leader, because across the 110 gate points a median of 21.6% of
the tile centroids inside the 80 m join fall within a building footprint, and the highest at any gate is
37%, so a gate-area figure is mostly an outdoor ramp reading (per concourse: Concourse A 20.6%,
Concourse B 22.9%, Concourse C 16.5%, Concourse D 20.1%, Concourse E 21.6%, Concourse F 26.6%). Per
carrier, the lowest building is the Gates D11 to D18 building for all three (T-Mobile 16%, Verizon 9%,
AT&T 57%), and the highest is Terminal 3 for T-Mobile (92%) and Terminal 4 for Verizon (84%) and AT&T
(74%). The leader changes from building to building, which is the pattern a phased in-building program
produces and equally the pattern macro-site geometry can produce; the attestation in Recommendation 1
decides which. Ookla's own Indoor layer is not used here (section 15).
How to read a tile map
Each tile is 0.0001° by 0.0001°, about 9 m by 11 m at 33.4° N, and is colored by the locked ladder from
the tile's aggregated RSRP over the window. A tile with no reading had no 5G sample from the Android
panel in the window. That means no measurement, not no coverage. Gate labels mark the pinned OSM gate
points.
Section 5
Tenant level
117 points · tile basis behind each category
The 117 mapped points come from an OpenStreetMap capture pinned 2026-09-16 (points only). Category point
counts sit below the 30-point floor. Category comparisons therefore run on the tile basis behind each
category's points: tiles within 80 m of any point, deduplicated. Point counts appear on every axis.
5.1 Category signal strength and quality
Ranked categories clear the floor for every carrier: transport_admin, retail, food, lounge, services.
Categories reported as counts only (a carrier's tile basis below 30): other (3 points). Carrier-to-carrier
spread per category on the tile median: transport_admin 4 dB, retail 1 dB, food 1 dB, lounge 1 dB, services
3 dB. Spreads at or under 1 dB are not rankings, and tile RSRP is integer-valued, so a category ordering
is not published here. Read the chart as a pattern; the point counts are small and every category pool
includes apron and roadway tiles. Figs. 20 and 21 (median 5G RSRP and SINR by tenant category) ship in the
PDF.
Periphery minus terminal median 5G RSRP, tile basis.
Terminal categories cover the concessions and services inside the terminal footprints; periphery
categories cover the parking structures, the Sky Train and transit points, and roadside businesses. A large
positive periphery-minus-terminal delta is consistent with macro spillover. A flat or negative delta is
ambiguous between close-in macro coverage and an in-building source. Distance to macro sites confounds both
readings, and the Site Finder layer (section 15) is queued to resolve it. Deltas: T-Mobile −1.0 dB,
Verizon +2.0 dB, AT&T −1.0 dB. Tile RSRP is integer-valued, so medians and deltas quantize to whole dB.
Fig. 22 · Median 5G RSRP · terminal vs periphery tenant categoriesSource · Ookla
Per-carrier median 5G RSRP, terminal tenant categories vs periphery categories, on the tile basis. Proxy: category-based masking, not Ookla's building-polygon indoor mask (queued, section 15). Bars rise from −115 dBm, so a longer bar is a stronger signal. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
5.3 Airline gate clusters
An airline reads its exposure through its concourse and the passenger's carrier. The concourse column is
the airline set the official gate map and the airport's airline list place there; a per-airline gate
assignment is not held for either terminal, so the rows are concourse groupings, not airline-specific
measurements. Read the medians with the margins shown: margins at or under 1 dB are not rankings, tile
RSRP is integer-valued, and the pool behind each row includes apron and roadway.
Concourse: airlines not independently verified
Gates (n)
T-Mobile median 5G RSRP
Verizon median 5G RSRP
AT&T median 5G RSRP
Concourse A (Terminal 4): American Airlines
28
−93.0 dBmn=1,957
−92.0 dBmn=1,865
−93.0 dBmn=927
Concourse B (Terminal 4): American Airlines (B23-B28 international)
30
−83.0 dBmn=2,492
−89.0 dBmn=1,818
−94.0 dBmn=886
Concourse C (Terminal 4): Southwest Airlines
16
−89.0 dBmn=1,258
−92.0 dBmn=1,172
−90.0 dBmn=480
Concourse D (Terminal 4): Southwest Airlines
16
−97.0 dBmn=994
−94.0 dBmn=915
−91.0 dBmn=440
Concourse E (Terminal 3): Air Canada, Alaska, Hawaiian, United (primarily)
10
−92.0 dBmn=1,267
−92.0 dBmn=1,052
−86.0 dBmn=505
Concourse F (Terminal 3): Delta and other Terminal 3 carriers (common use)
10
−94.0 dBmn=2,064
−89.0 dBmn=2,051
−94.0 dBmn=767
Tile basis; n = tiles per carrier per concourse. Airline assignments transcribed from the official gate map and the airport's airline list.
5.4 Retail and food brand coverage
Multi-location brands are a cross-zone consistency check. Point counts are small (6 at most), so this
table is a count-only view and is never ranked. Each point value is a mean over the tiles within 80 m of
the point (per-point tile n in Appendix A of the PDF).
Brand
Points (n)
T-Mobile median 5G RSRP
Verizon median 5G RSRP
AT&T median 5G RSRP
Hudson
1
−86.2 dBm
−89.0 dBm
−88.0 dBm
Starbucks
6
−86.8 dBm
−87.7 dBm
−88.5 dBm
Peet's
1
−93.8 dBm
−88.2 dBm
−89.3 dBm
Panda Express
1
−87.8 dBm
−92.3 dBm
−92.0 dBm
Wendy's
1
−87.4 dBm
−86.1 dBm
−88.7 dBm
Barrio Cafe
1
−96.7 dBm
−91.6 dBm
−93.5 dBm
Four Peaks
1
−96.5 dBm
−90.8 dBm
−95.1 dBm
Zinburger
1
−89.9 dBm
−88.9 dBm
−90.2 dBm
Blanco
1
−89.7 dBm
−88.8 dBm
−90.7 dBm
Shake Shack
1
−87.4 dBm
−92.0 dBm
−91.9 dBm
Cartel
1
−83.1 dBm
−87.9 dBm
−86.6 dBm
Tammie Coe
1
−96.2 dBm
−93.5 dBm
−95.6 dBm
Pei Wei
1
−96.6 dBm
−90.7 dBm
−95.2 dBm
La Grande Orange
1
−97.2 dBm
−91.2 dBm
−93.4 dBm
Humble Pie
2
−93.7 dBm
−89.4 dBm
−91.8 dBm
Matt's Big Breakfast
1
−88.0 dBm
−85.5 dBm
−90.9 dBm
InMotion
3
−87.1 dBm
−87.9 dBm
−88.1 dBm
Arizona Highways
1
−94.0 dBm
−87.7 dBm
−89.9 dBm
Count-only view (fewer than 30 points per brand); never ranked.
5.5 Airline lounges present in the OpenStreetMap capture
2 lounge points are in the pinned capture: Centurion Lounge and Escape Lounge. These are single-point
aggregates over the tiles within 80 m of each point, reported for scope and not ranked.
Lounge point
Zone
T-Mobile 5G RSRP (class, n tiles)
Verizon 5G RSRP (class, n tiles)
AT&T 5G RSRP (class, n tiles)
Centurion Lounge
Concourse B
−86.3 dBmhealthy, n=177
−89.6 dBmhealthy, n=159
−89.7 dBmhealthy, n=168
Escape Lounge
Concourse B
−86.3 dBmhealthy, n=173
−89.7 dBmhealthy, n=155
−90.2 dBmhealthy, n=164
Section 6
Load and density
Unique devices · zone level
Density is this dataset's demand signal. Ookla's User_Density field counts the unique Android devices
observed per tile over the window, deduplicated. It does not count simultaneous users and it does not
measure time spent.
Zone
Median devices per tile
p95
Usage tiles (n)
Worst carrier · below healthy 5G
Concourse A (T4)
5
77
2,647
AT&T · 42% of tiles
Concourse B (T4)
3
38
3,308
AT&T
Concourse C (T4)
2
46
1,984
Verizon
Concourse D (T4)
3
48
1,559
T-Mobile
Concourse E (T3)
2
61
1,932
T-Mobile
Concourse F (T3)
2
42
3,421
AT&T and T-Mobile tied
Venue (on-airport)
2
n/a
26,948
n/a
User_Density = unique devices over the window, deduplicated. Tile basis. Density-weighted ranking = worst-carrier below-healthy 5G tile share × median density. A below-healthy margin under 1 point is a tie.
The density-weighted ranking multiplies each zone's worst-carrier below-healthy 5G tile share by its
median density, so it points to where weak signal meets the most devices. Concourse A leads the ranking
(AT&T below healthy on 42% of tiles, median density 5). The on-airport median is 2 devices per tile
across 26,948 usage tiles.
Fig. 23 · Unique-device density by concourse zone · density-weighted weak-share rankingSource · Ookla
Unique-device density by concourse zone (median and p95 per tile) and the density-weighted weak-share ranking (tile basis). User_Density = unique devices over the window, deduplicated. Source: Ookla Cell Analytics All Networks usage export, 2025-09-01 to 2026-08-31.
Fig. 24 · Observed data volume by concourse zone and bearerSource · Ookla
Observed data volume by concourse zone and bearer (GB over the 12-month window). Source: Ookla Cell Analytics All Networks usage export, 2025-09-01 to 2026-08-31.
Usage volume shows where bandwidth is consumed, by bearer and by zone. It is observed sample volume from
the crowdsource panel. It is not total traffic.
Section 7
Wi-Fi as an augmentation layer
Venue and zone level only
Wi-Fi presence (a different figure from the offload share): Wi-Fi usage is observed on 60.0% of the 19,449
on-airport usage tiles that report the Wi-Fi field. A non-reporting tile means no measurement and is
excluded from the denominator; it is not evidence of no Wi-Fi. The augmentation opportunity is largest
where the worst-served carrier is weakest and Wi-Fi presence is high. Fig. 25 (worst-carrier healthy share
vs Wi-Fi presence by zone) ships in the PDF.
7.1 Offload share (venue and zone level only)
On tiles assigned to a concourse zone, 46.5% of observed data volume moves over Wi-Fi (Σ Wi-Fi MB ÷
Σ all MB on the All Networks panel, n=14,851 tiles); over the whole clip the figure is 47.1% (n=26,948).
Scope decomposition: the 12,097 usage tiles outside every zone run 50.2% Wi-Fi. The 200 m buffer around
the aerodrome captures the Sky Train stations, the Rental Car Center approach, the economy parking lots, and
roadway; those are not concourse tiles, which is why the concourse-zone figure is the one on the cover.
Fig. 26 · Wi-Fi share of observed data volume by concourse zoneSource · Ookla
Wi-Fi share of observed data volume by concourse zone, with the venue reference line. Source: Ookla Cell Analytics All Networks usage export, 2025-09-01 to 2026-08-31.
No per-carrier offload
Locked rule (section 16): offload is computed at venue and zone level only. The export does not attribute
Wi-Fi usage to any cellular carrier's subscribers, so per-carrier offload is not derivable and not
implied. The figure blends the public network with tenant, airline lounge, hotel, and personal
networks.
7.2 Tenant-level Wi-Fi footprint
Wi-Fi presence and observed volume peak where passengers spend time. The public network is
Free PHX Boingo WiFi, operated by Boingo Wireless under City of Phoenix contract 152010, a revenue
contract covering the public areas of both terminals and the Rental Car Center; the City extended it
through July 31, 2026 and then month to month while it transitions the service in house
held. The Wi-Fi figures here come solely from the Ookla usage layer.
Controller telemetry (section 15) would convert presence into capacity. Fig. 27 (Wi-Fi observation rate and
intensity across tenant categories) ships in the PDF.
Section 8
Private wireless
CBRS · operational workloads · open items
No public record of a Citizens Broadband Radio Service, private LTE or private 5G network at the airport
was located; Phoenix public safety radio runs on the Regional Wireless Cooperative P25 system, and a 2011
trade report describes a distributed antenna system built for in-building radio coverage at the airport
not independently verified. The Ookla drop for this window also contained a separate
three-tile export labelled with a CBRS operator; three tiles is below the 30-tile floor and it is held in
the audit dossier as a count, not surfaced here. The public Ookla layer does not characterize private
networks. Whether a Citizens Broadband Radio Service (CBRS) private-network layer is scoped alongside the
City DAS program is a design question for the Aviation Department. AT&T's FirstNet Band 14 (LTE700PS)
appears on 335 AT&T LTE tiles (section 9.1).
The two open items
Two items about the in-building layer are not held at this issue, and both are requests to the Aviation
Department rather than gaps in the measurement. First, the entity awarded solicitation
AVN RCS 21-008: no City Council award record, press release or vendor announcement naming the awarded
entity was found. Second, the program's per-building phase map, which would say which concourses
carried service on the new system during this window. Recommendation 1 asks for both.
Everything measured in this report stands without them.
Typical large-hub private-network workloads:
Workload
Why a private network fits
Coordination required
Baggage handling system instrumentation
Deterministic latency, contained RF, isolation from passenger traffic
Aviation Department, baggage system vendors
Ramp and ground-service vehicle telematics
Outdoor airside coverage with controlled SIM provisioning
Airside operations
Push-to-talk and operational voice
Guaranteed-quality voice independent of public carriers
Phoenix Police Airport Bureau, ground handlers
Fixed cameras and IoT in non-public areas
Bandwidth without consuming public-carrier capacity
Facilities maintenance
Construction-phase connectivity (Terminal 3 and Terminal 4 program work)
Coverage that survives phased construction sequencing
Program management, contractors
Section 9
Spectrum and service availability
Bands · availability
9.1 Band distribution
Legend: Ookla band labels are technology plus approximate frequency in MHz, printed verbatim from the
export. For example, NR2500 and NR2600 are both 2.5 GHz spectrum, and LTE800 and NR850 are both 850 MHz
cellular spectrum. LTE700PS is FirstNet Band 14, the public-safety broadband band. Cell, PCI, and band
fields populate only where Ookla resolved a serving cell, so band n is smaller than RSRP n (section 16).
The bands a tile attaches to most often show which spectrum each carrier puts to work at the venue.
Sub-1 GHz attachment implies coverage-grade bands. Mid-band attachment implies strong penetration or an
in-building source. AT&T's 5G band mix rests on 1,366 cell-resolved tiles against its 6,179 5G-reporting
tiles. LTE1800 rows are an Ookla band label as exported, not a United States band allocation. Fig. 28
(strongest-serving band distribution per carrier) ships in the PDF.
Carrier
5G NR strongest band · tiles (n) · share of NR footprint
LTE strongest band · tiles (n) · share of LTE footprint
Labels are Ookla's, printed verbatim. Share = band tiles ÷ the carrier's band-reporting tile footprint (PDF Appendix B).
9.2 Service availability
Service availability is near-universal for all three carriers and is stated here rather than charted,
because every value sits within three points of 100 and a chart of that adds nothing. On tiles that report
the field, mean own-network availability is T-Mobile 99%, Verizon 97%, AT&T 98%, and mean general-LTE
availability is T-Mobile 99%, Verizon 100%, AT&T 100%. That is the graceful-degradation floor: a device
keeps a usable LTE layer almost everywhere on the campus. Basis disclosure: these fields are not
populated on every tile. The own-network mean rests on T-Mobile 15,697, Verizon 16,079, AT&T 10,525 tiles
and the general-LTE mean on T-Mobile 15,719, Verizon 16,088, AT&T 10,530 tiles, which is T-Mobile 83%,
Verizon 87%, AT&T 82% of each carrier's on-airport tiles; the field-population table in section 16 lists
every field on the same basis.
Section 10
Carrier-asymmetry atlas
Worst-carrier tally per gate
The top 20 gates by best-minus-worst carrier 5G RSRP spread are where a passenger's experience varies
most by carrier (spreads of 9 to 26 dB). The best carrier across the top 20 is T-Mobile or Verizon.
Worst-carrier tally across all 110 ranked gates: AT&T 49, Verizon 33, T-Mobile 28. The top five sit
in Concourse B (3) and Concourse F (2). They serve two purposes: worst-case examples for the stakeholder
narrative, and the validation-point list for post-deployment monitoring (section 14). A multi-carrier
in-building system should compress this spread.
Gate (OSM ref)
Concourse · airlines not independently verified
Worst carrier this window
Spread this window
B·B15B
Concourse B · American Airlines (B23-B28 international)
AT&T
25.9 dB
B·B15A
Concourse B · American Airlines (B23-B28 international)
AT&T
20.1 dB
F·F1
Concourse F · Delta and other Terminal 3 carriers (common use)
AT&T
15.5 dB
F·F2
Concourse F · Delta and other Terminal 3 carriers (common use)
AT&T
14.5 dB
B·B1A
Concourse B · American Airlines (B23-B28 international)
AT&T
13.4 dB
Spread = best minus worst carrier mean 5G RSRP at the gate. Among the 110 gates where every carrier's per-gate n is at least 30 tiles. B1A, B15A and B15B are OpenStreetMap hardstand points that are not on the published gate list (section 2.1).
Fig. 29 · Carrier-asymmetry atlas · top 20 gates by spreadSource · Ookla
Top 20 gates by best-minus-worst carrier 5G RSRP spread, among the 110 gates where every carrier's per-gate n is at least 30 tiles. Bar color is the worst carrier at that gate. Gate refs follow the OpenStreetMap inventory. Source: Ookla Cell Analytics per-carrier RF tile export, 2025-09-01 to 2026-08-31.
Fig. 30 · Worst-carrier 5G class per tile and unique-device density per tileSource · Ookla
Left: each terminal-area tile colored by the worst carrier's 5G NR class where all three carriers report. Right: User_Density per usage tile (unique devices, deduplicated) on a log scale over the same extent. Source: Ookla Cell Analytics per-carrier RF tile export and All Networks usage export, 2025-09-01 to 2026-08-31.
Fig. 30 pairs the two questions an operator asks first: where is the weakest carrier weak, and where are
the devices. Amber concentrates over the east end of Terminal 4 and the Terminal 3 concourses, and the
density panel shows the concourse cores and the Sky Train stations carrying the highest unique-device
counts.
Section 11
Federal and operational-critical tenants
Single-point aggregates · scope only
1 operational-critical point is identifiable by name in the pinned capture (Phoenix Police Airport
Bureau); no other federal or operational-critical point is identifiable by name in the capture. AT&T's
FirstNet Band 14 (LTE700PS) appears on 335 AT&T LTE tiles (section 9.1); public-safety coverage posture is
a section 15 layer. These readings are single-point aggregates, reported for scope and not ranked.
One point on crowdsourced tiles is not a coverage assessment. Recommendation 5 asks for a walk test here.
Photo · Terminal 4 international arrivals baggage claim, December 2024Wikimedia Commons
U.S. Customs and Border Protection officers in the Terminal 4 international arrivals baggage claim, December 2024. Photo: CBP Photography, Public domain, via Wikimedia Commons. Illustrative, resized only.
Section 12
Share of gate points below the healthy line, and why passenger exposure is not derivable
Illustrative proxy · not a measurement
Translation method: PHX recorded 25,595,723 revenue enplanements in calendar 2024 (FAA
not independently verified) and 51,618,649 total passengers in 2025 (airport statistics
page, both directions held); the two bases are not interchangeable and only
the FAA enplanement figure belongs in any translation. Per-gate volumes are not held, so the only
translation this report offers is illustrative and stays in prose: distributing annual enplanements
uniformly across the 110 measured gates would place the share of gates below healthy 5G onto the same
share of enplanements.
Carrier
Gates below healthy 5G
Share of gates
Gates within 1 dB of the line (sensitivity)
T-Mobile
29 of 110
26.4%
10
Verizon
17 of 110
15.5%
6
AT&T
22 of 110
20.0%
16
Illustrative proxy, kept in prose
It is not a measurement, it assigns every gate the same volume regardless of airline, and it is
not tabulated for that reason. A gate-level passenger feed (section 15) would replace it. The point
stands without it: T-Mobile 26.4%, Verizon 15.5% and AT&T 20.0% of gates sit below the healthy 5G
threshold in this window, with T-Mobile 10, Verizon 6 and AT&T 16 of those gates within 1 dB of the
line.
Section 13
Prioritized recommendations
Ordered by dependency · each with an owner
Obtain the two items the public record does not hold
The entity awarded solicitation AVN RCS 21-008, and the DAS program's per-building phase map, with the
bands (LTE, 5G NR) each of AT&T, Verizon and T-Mobile carries per building and the go-live schedule
through spring 2027. The measured pattern, a different least-weak carrier over each of the Terminal 3,
Terminal 4 and Gates D11 to D18 footprints (section 4.7), is consistent with more than one explanation,
and the answer decides what the after reading is measured against.
Owner · City of Phoenix Aviation Department with the DAS program team
Adopt this window as the baseline reading and re-measure per building as each goes live
Section 14 lists the per-carrier, per-zone criteria. Take the weakest 5 carrier-zone cells first, all
at or below 58% healthy on the tile basis: T-Mobile in Concourse D (44% of 994 tiles); Verizon in
Concourse D (54% of 915 tiles); AT&T in Concourse F (57% of 767 tiles); AT&T in Concourse B (58% of
886 tiles); T-Mobile in Concourse F (58% of 2,064 tiles). They span Terminal 3 (2) and Terminal 4 (3),
so those concourses carry the largest measured gap and should be sequenced first where the program
schedule allows.
Owner · Imagine Wireless with the Aviation Department
Make the Gates D11 to D18 footprint the first place the programme is re-measured
For all three carriers: it is the weakest building measured (T-Mobile 16%, Verizon 9%, AT&T 57% of
tiles healthy, medians T-Mobile −102.0, Verizon −106.0, AT&T −94.0 dBm). The criteria in section 14 are
indicative monitoring targets for a like-for-like re-measurement. They are NOT acceptance criteria and
must not be written into a carrier agreement or a DAS acceptance document: acceptance is per-carrier
commissioning by walk test and scanner survey.
Owner · Aviation Department with the DAS program team and AT&T, Verizon and T-Mobile
Treat Wi-Fi as a coexisting capacity layer in the design
46% of observed data volume in the concourse zones rides Wi-Fi (section 7.1); that figure blends the
public network with tenant, lounge, hotel and personal Wi-Fi and is not attributable to any one
network. Obtain controller telemetry for the public network to size it.
Owner · City of Phoenix Aviation Department, Wi-Fi operator
Verify public-safety coverage at the Phoenix Police Airport Bureau point by walk test
Across the three carriers the 5G readings at that single point span 21 dB (67 tiles per carrier), and
the weakest of the three is AT&T, which is the FirstNet carrier. One point on crowdsourced tiles is not
a coverage assessment, which is why this is a walk test rather than a finding, and emergency responder
radio coverage is governed by the authority having jurisdiction and NFPA 1221, not by this
instrument.
Owner · Aviation Department with the Phoenix Police Airport Bureau and the authority having jurisdiction
Decide the CBRS and private-wireless question alongside the DAS program
Decide it alongside the program (section 8), not after it, so the DAS coverage exhibit's
decommissioning clause for standalone systems is applied with the private-network plan in view.
Owner · City of Phoenix Aviation Department
Section 14
Post-deployment monitoring criteria and re-measurement plan
Indicative · baseline-and-after · locked ladder
Indicative monitoring criteria, not acceptance criteria
These are indicative monitoring criteria for a like-for-like re-measurement. They are not the
acceptance test for any system, they are not contractual, and they must not be written into a carrier
agreement or a DAS acceptance document: acceptance is per-carrier commissioning by walk test and
scanner survey. Every threshold below is the locked ladder from section 16 (−95.0 dBm healthy) plus this
window's measured baseline recorded for comparison. This report sets no target share: a share to
aim at depends on the design, and on window-to-window variation this single window cannot estimate. The
baseline itself may blend served and unserved space, because which buildings carried service on the new
system during the window is not held (section 8).
Gate refs and counts come from the OpenStreetMap inventory (Concourse A 28, Concourse B 30, Concourse C
16, Concourse D 16, Concourse E 10, Concourse F 10). Gate-named criteria must be reconciled against the
Aviation Department's gate plan before adoption (section 2.1).
Zone, building, or point
Carrier
Criterion (threshold)
Re-test
All measured gates
Every hosted carrier
5G SS-RSRP median ≥ −95 dBm per gate
Like-for-like window after go-live
All measured gate points
T-Mobile
Healthy 5G share above this window's baseline of 74% (81 of 110 gate points); no target share is set here
Like-for-like window after go-live
Concourse A
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 60% (n=1,957)
Like-for-like window after the zone's building goes live
Concourse B
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 92% (n=2,492)
Like-for-like window after the zone's building goes live
Concourse C
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 77% (n=1,258)
Like-for-like window after the zone's building goes live
Concourse D
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 44% (n=994)
Like-for-like window after the zone's building goes live
Concourse E
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 67% (n=1,267)
Like-for-like window after the zone's building goes live
Concourse F
T-Mobile
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 58% (n=2,064)
Like-for-like window after the zone's building goes live
T3: Terminal 3
T-Mobile
Median over the footprint at or above −95.0 dBm (locked ladder; this window −86.0 dBm); healthy share compared against this window's baseline of 92% (n=344)
Like-for-like window after that building goes live
T4: Terminal 4
T-Mobile
Median over the footprint at or above −95.0 dBm (locked ladder; this window −90.0 dBm); healthy share compared against this window's baseline of 77% (n=546)
Like-for-like window after that building goes live
T4: Gates D11-D18
T-Mobile
Median over the footprint at or above −95.0 dBm (locked ladder; this window −102.0 dBm); healthy share compared against this window's baseline of 16% (n=99)
Like-for-like window after that building goes live
All measured gate points
Verizon
Healthy 5G share above this window's baseline of 85% (93 of 110 gate points); no target share is set here
Like-for-like window after go-live
Concourse A
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 63% (n=1,865)
Like-for-like window after the zone's building goes live
Concourse B
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 78% (n=1,818)
Like-for-like window after the zone's building goes live
Concourse C
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 66% (n=1,172)
Like-for-like window after the zone's building goes live
Concourse D
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 54% (n=915)
Like-for-like window after the zone's building goes live
Concourse E
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 72% (n=1,052)
Like-for-like window after the zone's building goes live
Concourse F
Verizon
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 75% (n=2,051)
Like-for-like window after the zone's building goes live
T3: Terminal 3
Verizon
Median over the footprint at or above −95.0 dBm (locked ladder; this window −90.0 dBm); healthy share compared against this window's baseline of 82% (n=333)
Like-for-like window after that building goes live
T4: Terminal 4
Verizon
Median over the footprint at or above −95.0 dBm (locked ladder; this window −89.0 dBm); healthy share compared against this window's baseline of 84% (n=540)
Like-for-like window after that building goes live
T4: Gates D11-D18
Verizon
Median over the footprint at or above −95.0 dBm (locked ladder; this window −106.0 dBm); healthy share compared against this window's baseline of 9% (n=98)
Like-for-like window after that building goes live
All measured gate points
AT&T
Healthy 5G share above this window's baseline of 80% (88 of 110 gate points); no target share is set here
Like-for-like window after go-live
Concourse A
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 58% (n=927)
Like-for-like window after the zone's building goes live
Concourse B
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 58% (n=886)
Like-for-like window after the zone's building goes live
Concourse C
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 71% (n=480)
Like-for-like window after the zone's building goes live
Concourse D
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 67% (n=440)
Like-for-like window after the zone's building goes live
Concourse E
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 73% (n=505)
Like-for-like window after the zone's building goes live
Concourse F
AT&T
Zone tile median at or above −95.0 dBm (locked ladder); healthy share compared against this window's baseline of 57% (n=767)
Like-for-like window after the zone's building goes live
T3: Terminal 3
AT&T
Median over the footprint at or above −95.0 dBm (locked ladder; this window −90.0 dBm); healthy share compared against this window's baseline of 69% (n=227)
Like-for-like window after that building goes live
T4: Terminal 4
AT&T
Median over the footprint at or above −95.0 dBm (locked ladder; this window −91.0 dBm); healthy share compared against this window's baseline of 74% (n=412)
Like-for-like window after that building goes live
T4: Gates D11-D18
AT&T
Median over the footprint at or above −95.0 dBm (locked ladder; this window −94.0 dBm); healthy share compared against this window's baseline of 57% (n=68)
Like-for-like window after that building goes live
Terminal 4 and Terminal 3
T-Mobile
LTE tile median ≥ −95 dBm per zone (this window: Concourse A −98.0; Concourse B −87.0; Concourse C −92.0; Concourse D −102.0; Concourse E −90.0; Concourse F −93.0 dBm)
Like-for-like window after go-live
Gate B·B15B (OSM ref)
worst this window: AT&T
Carrier-to-carrier spread compared against this window's 25.9 dB; a multi-carrier in-building source should compress it
Like-for-like window after go-live
Gate B·B15A (OSM ref)
worst this window: AT&T
Carrier-to-carrier spread compared against this window's 20.1 dB; a multi-carrier in-building source should compress it
Like-for-like window after go-live
Gate F·F1 (OSM ref)
worst this window: AT&T
Carrier-to-carrier spread compared against this window's 15.5 dB; a multi-carrier in-building source should compress it
Like-for-like window after go-live
Gate F·F2 (OSM ref)
worst this window: AT&T
Carrier-to-carrier spread compared against this window's 14.5 dB; a multi-carrier in-building source should compress it
Like-for-like window after go-live
Gate B·B1A (OSM ref)
worst this window: AT&T
Carrier-to-carrier spread compared against this window's 13.4 dB; a multi-carrier in-building source should compress it
Like-for-like window after go-live
Re-measurement uses the same instrument as this baseline (Ookla Cell Analytics, identical thresholds).
The window is like for like: the same 12-month length, or a matched shorter window that is disclosed
alongside the result. Ookla's date-range comparison renders the before and after directly. Imagine
Wireless re-measures over that window after any in-building change goes live; the crowdsourced tiles serve
as the baseline-and-after comparison and acceptance itself is per-carrier commissioning.
Section 15
Next layers to add
Queued for the next revision
Layer
What it resolves
Source
Ookla Site Finder centroids (LTE and 5G)
Macro-site geometry behind the section 3 and 5.2 readings; this issue publishes no site figure, so those readings rest on the RF tiles alone; queued for the next issue
Ookla portal pull (Imagine Wireless)
DAS program inventory
The awarded operator of solicitation AVN RCS 21-008, the per-building phase map, the bands each carrier carries per building, and the go-live schedule (Recommendation 1)
City of Phoenix Aviation Department, DAS program team
Google Places tenant capture
Adds admin, services, and lounge points to the OpenStreetMap list and thickens the category cells
Google Places (New) pull (Imagine Wireless)
Indoor mask
OSM building-polygon mask shipped in this issue (section 4.7); Ookla native Indoor layer to be pinned as portal captures
Ookla portal captures (Imagine Wireless)
Date-range comparison
Before and after for each building as the DAS goes live; converts this snapshot into a trend
Ookla portal pull
Poor multi-network coverage
Vendor-ranked building-versus-carrier problem split; corroborates section 4.3
Ookla portal pull
3D Z-axis view
15 m vertical bands; the export is flat but the portal is not
Ookla portal pull
Per-gate passenger counts
Replaces the section 12 uniform-distribution proxy with measured exposure
Aviation Department, airline schedules
Gate plan reconciliation
Aligns OpenStreetMap gate refs with the Aviation Department's gate list before gate-named criteria are adopted
City of Phoenix Aviation Department
Wi-Fi controller telemetry
Airtime, retries, client density; converts presence into capacity
Wi-Fi operator, Aviation Department
Contract records for PHX
The knowledge graph's contracts query was not available at this issue; the DAS revenue contract's award date, term, and per-building scope
Imagine knowledge graph; City of Phoenix procurement record
Public-safety (NFPA 1221) coverage
Emergency responder radio coverage posture; FirstNet Band 14 is present in the data
Authority having jurisdiction, integrator
Throughput and latency series
Section 1 engages with Ookla's published airport benchmark for PHX; the underlying Speedtest series per terminal and per carrier would pair throughput with the signal readings in this report, which carries neither
Ookla portal pull (Imagine Wireless)
Walk test: jet bridges, baggage corridors, construction zones
Areas crowdsource cannot reach; the acceptance instrument
Scanner survey
Section 16
Method, thresholds, confidence, and caveats
Thresholds · provenance · reproducibility
Rule
Value
Measurement source
Ookla Cell Analytics per-carrier RF tile exports and the All Networks usage export for PHX, 12-month window 2025-09-01 through 2026-08-31, clipped to the KPHX aerodrome polygon (OpenStreetMap relation 8367318, largest outer ring) plus a 200 m buffer, intersected with the single export rectangle shared by all four files (latitude 33.422 to 33.4523, longitude -112.0422 to -111.9718). 50,313 on-airport RF tiles are retained from 111,616 exported; 26,948 usage tiles from 61,367. The rectangle contains the whole terminal area; only the buffered strip south of latitude 33.422 (the south side of runway 7R/25L, no terminal, no gate) lies outside it. Measurement freshness at issue is 16 days, within the platform's 90-day policy.
Tile geometry and sample basis
Each RF tile is 0.0001° by 0.0001°, about 9 m by 11 m at 33.4° N. The three carrier panels cover 27,206 unique tile centroids on the airport, 8,517 of them seen by all three carriers. The export carries no per-tile scan count, so every n in this report counts tiles, not scans. The 30-tile floor guards spatial support, not scan volume. Ookla aggregates each tile from its panel over the window.
Joins and zones
Tiles to gate and tenant points within an 80 m radius. Gates and gate refs come from a pinned OpenStreetMap inventory (110 gates: Concourse A 28, Concourse B 30, Concourse C 16, Concourse D 16, Concourse E 10, Concourse F 10; the concourse letter in each ref sets the zone, a ref must carry a digit, and the build checks every lettered gate against the pinned terminal footprint with a 150 m nearest-footprint fallback). The official Terminal 4 Level 3 map, revision October 2, 2022, publishes gates A1 to A30, B1 to B28 with B23 to B28 international, C1 to C9 and C11 to C19, D1 to D8, and the airport's history page adds gates D11 to D18 in the 2022 concourse; Terminal 3 publishes gates E1 to E10 in the North Concourse and F1 to F15 in the South Concourse, with the terminal total stated as 25 gates in the 2020 modernization record and as 20 in a 2025 City release that adds six more held, so gate-named criteria must be reconciled against the Aviation Department's gate plan before adoption (the reconciliation is tabulated in section 2.1). Tenants come from a pinned OpenStreetMap capture (117 points, no polygons, no Z-axis). Tile-to-zone assignment: nearest gate or tenant point within 500 m (disclosed proxy). Tenant zone attribution: the concourse of the nearest lettered gate within 400 m, and the terminal by point-in-polygon against the same footprints (disclosed proxy).
Building mask and maps
Building mask (section 4.7): point-in-polygon of tile centroids against the pinned OSM footprints (Terminal 3, Terminal 4, Gates D11 to D18), so it selects tiles over a building, not tiles measured inside it. Maps: equirectangular axes in degrees, north up, no basemap tiles, rebuilt from the pinned CSVs.
Central tendency and carrier order
Median for every headline figure; means only where labeled. Carrier mapping: the export's T-Mobile US, Verizon, and AT&T US operators are printed as T-Mobile, Verizon, and AT&T. Carrier ordering throughout: strongest at the gates first (T-Mobile and Verizon, which are tied, then AT&T), asserted against the data at build time.
Threshold ladder (locked)
5G NR SS-RSRP and LTE RSRP healthy ≥ −95 dBm, at risk −95 to −110, critical < −110. 5G SS-SINR healthy ≥ 10 dB, at risk 0 to 10, critical < 0. RSRQ healthy ≥ −10 dB. No venue-specified threshold was provided. Noise floor: margins under 1 dB are not rankings; every rank statement in this report prints its margin, and the T-Mobile-to-Verizon gate-median margin of 0.5 dB is reported as a tie.
Minimum-sample rule
No chart or comparative claim ships on fewer than 30 underlying tiles or points. Only Concourse B clears the gate floor (Concourse A 28, Concourse C 16, Concourse D 16, Concourse E 10, Concourse F 10 gates do not), Terminal 3 holds 20 lettered gates, and every tenant category holds fewer than 30 points, so those tallies appear as counts with n visible and are not ranked; zone, terminal, and category comparisons run on the tile basis for one consistent basis (applied: sections 4.3 to 4.7, 5.4, 5.5, 10, 11).
Confounded statistics, not compared across carriers
Two statistics on this page mix network state with the measuring panel and are never compared across carriers. The 5G footprint share (section 3.4) counts tiles reporting a 5G reading, and a tile reports 5G only if a handset in the panel could report that carrier's 5G band, so the share mixes deployment with each carrier's mix of 5G-capable Android handsets. Cross-carrier SINR (section 3) depends on the serving band's bandwidth and on neighbour loading, and the three carriers sit on different bands here, so a cross-carrier SINR gap is not like-for-like; within a carrier, SINR across zones is a fair comparison and is used that way. PCI diversity (section 4.2) is drawn per 1,000 cell-resolved tiles because a unique count rises with sample size, and the cell-resolved samples differ by more than an order of magnitude.
Units rule
Ookla User_Density is the count of unique Android devices observed per tile over the measurement window, deduplicated. It does not count simultaneous users and it does not measure time spent, and neither reading appears anywhere in this report. Wi-Fi offload = Σ Wi-Fi MB ÷ Σ (Wi-Fi + Mobile) MB by volume, computed at venue and zone level only on the All Networks panel. Per-carrier offload is not derivable from this dataset and is not implied. The offload figure cannot separate the public network from tenant or personal networks.
Sample basis and inference limits
Crowdsourced background scans from consumer Android handsets (a panel Ookla characterizes as about 3% of devices; iOS devices are not sampled), aggregated per tile over 12 months. The panel's composition differs by carrier, so cross-carrier comparisons inherit a device-population bias this export cannot correct. Supported: medians, shares, and distributions at venue, zone, gate-cluster, and tenant-category level with n disclosed. Not supported: throughput, latency, or jitter (not in this export), per-carrier Wi-Fi offload, sub-tile positioning, and floor separation. Nothing in this report is measured indoors: there is no indoor mask beyond the building-polygon selection and no Z-axis, and gate values are tiles within 80 m of a point that can straddle the apron: the measured interior share of that join is a median of 21.6% across the 110 gate points (highest 37%), which is why terminal and building leaders in this report are named only on the building-polygon basis and gate-area results are reported as counts. The window overlaps the phased installation of the City DAS program, so any building that went live inside the window blends before and after readings.
Window and freshness
2025-09-01 to 2026-08-31 · issued 2026-09-16 · 16 days at issue, within the 90-day policy.
Confidence tagging
An entity fact whose source document is pinned in this build and checksummed in the verification bundle is tagged "held" inline; an entity fact this build cites without holding a copy is tagged "not independently verified" inline. Measured claims reproduce from the raw exports through the audit dossier that accompanies this report, which also records capture dates for transcribed values (concourse airline assignments) and the public source behind every entity statement. Facts from the Imagine knowledge graph are pinned in the build's source folder and tagged "Imagine knowledge graph" inline; the graph's contracts query was not available at this issue and it holds no leadership rows for PHX. The 90-day freshness policy is Imagine Wireless's reporting policy.
Population key
For the several n printed for one carrier: all on-airport tiles (T-Mobile 18,918, Verizon 18,513, AT&T 12,882); tiles inside the terminal extent drawn on the maps (T-Mobile 7,766, Verizon 7,966, AT&T 6,106); 5G-reporting tiles (T-Mobile 17,312, Verizon 14,734, AT&T 6,179); 5G-reporting tiles inside the map extent (T-Mobile 7,210, Verizon 6,844, AT&T 3,365). Cell, PCI, and band fields are smaller still (field population table below).
Caveats
The dataset is gate-area and public-corridor weighted; it does not cover jet-bridge interiors or baggage corridors. PCI diversity, serving-cell volatility, and band distributions are aggregate proxies, not active traces or carrier network-management data. The optimization priority index is Ookla's vendor-defined layer, reported as such. Engineering inferences about cause are labeled as inferences where they appear.
Provenance discipline
Every numerical value in this report is computed from the cited source files at build time, or carries an inline "held" tag when its source document is pinned in the build and checksummed in the verification bundle, or an inline "not independently verified" tag when this build cites a public record without holding a copy. No value is estimated or invented. Photographs are illustrative material from Wikimedia Commons under the licenses listed in the audit dossier and in each caption; they are not measurements.
Field population
Cell, PCI, and band fields populate only where Ookla resolved a serving cell, so their n is smaller than
the RSRP n for the same carrier. The table shows the number of on-airport tiles that populate each
field.
Field
T-Mobile
Verizon
AT&T
NR_SS_RSRP
17,312
14,734
6,179
NR_Strongest_Cell
15,874
9,793
1,365
NR_Strongest_PCI
15,875
9,798
1,366
NR_Strongest_Band
15,875
9,798
1,366
LTE_RSRP
10,861
15,151
12,209
LTE_Strongest_Cell
10,872
15,156
12,214
LTE_CQI
4,222
3,875
4,726
Optimization_Priority
8,310
13,655
10,462
General_LTE_Service
15,719
16,088
10,530
All on-airport tiles
18,918
18,513
12,882
Provenance tag legend
Tag
Meaning
held
An entity fact whose source document is pinned in this build and checksummed in the verification bundle: the City of Phoenix Aviation Department page, the airport's history, facts, statistics and airline pages, the official Terminal 4 Level 3 gate map, the Terminal 3 modernization media kit, the City's cellular DAS project page, the airport Wi-Fi page, and a Phoenix Aviation Advisory Board agenda packet. It is a held fact, never a measurement.
not independently verified
Public record this build cites without holding a copy: FAA statistics, award studies, capital-program releases, and trade press. It may corroborate a measured finding and never originates one.
Imagine knowledge graph
A fact from the pinned Imagine knowledge-graph snapshot, held in the build's source folder. The graph's contracts query was not available at this issue and it holds no leadership rows for PHX. The graph's DAS-operator line (a carrier consortium led by AT&T Mobility, 2025 to 2035) is reported as a graph record and is not asserted as the award answer.
transcribed
Transcribed values with a capture date recorded in the audit dossier: concourse airline assignments, the OpenStreetMap gate and tenant capture (pinned 2026-09-16), and every photograph.
Untagged numbers
Measured. Computed from the Ookla exports at build time and reproducible from the raw exports through the audit dossier that accompanies this report.
Photo credits
Photograph
Author
License
Date
Terminal 4 gate hold room
FASTILY
CC BY-SA 4.0
2023-08-11
Aerial view of Sky Harbor
Designism
CC0
2025-05-08
PHX Sky Train at the Terminal 4 station
Hunter Trick (TrickHunter)
CC BY-SA 4.0
2023-10-14
Southwest Airlines 737 on the Terminal 4 apron
BeeboMan
CC BY 4.0
2025-07-21
American Airlines 737 at Terminal 4 gate B22
Designism
CC0
2025-05-08
Terminal 4 concourse dining area
FASTILY
CC BY-SA 4.0
2023-08-11
Alaska Air Group aircraft at Terminal 3
Ziongarage
CC0
2025-04-27
Terminal 4 international arrivals baggage claim
CBP Photography
Public domain
2024-12-11
All photographs via Wikimedia Commons; used unmodified except for resizing. Every photograph is captured 2023 or later. Photographs are illustrative material and are not measurements.
Audit trail
Every measured value on this page is transcribed from the rev-1 report, whose numbers compute from the
pinned Ookla exports at build time and reproduce through the verification bundle. The audit dossier
(lane-tagged claims ledger, methodology, provenance manifest, findings log) ships with the bundle.
Deliverable subject to the two-lane human review gate; sign-off status accompanies this issue.
