Inside the terminal footprints (building-masked basis: 5G-reporting tiles inside the OSM terminal polygon, healthy share of those, median), T-Mobile leads 5 of 5 ranked terminals: Terminal 1, Terminal 4, Terminal 5, Terminal 7, Terminal 8. At the venue gate-area median the three carriers tie, all below the healthy line, and 75 of 134 gate points read weak for all three. Verizon is best-5G at 16 of 16 Terminal 1 gate points and healthy at 14 of 16 on the gate-area basis (counts, below the 30-gate floor; mean interior share 24%); the two bases describe different parts of Terminal 1 and are both reported.

Gate-area basis: mean of tiles within 80 m of the gate point, which straddles the apron. Adjacent margins 1.0 and 0.8 dB; the smaller is under the 1 dB noise floor, so the venue order is a tie.

Tied at the venue gate-area median; leads inside 0 of 5 terminal footprints (building-masked); best-5G at the most gate points in Terminal 1, Terminal 4 and Terminal 8 (gate-area counts). Venue gate-area median −98.3 dBm, healthy at 24 of 134 gate points (18%), with a 5G signal-to-interference-plus-noise ratio (SINR) median of +8.5 dB (45 of 134 healthy). It is best-5G at 54 of 110 decided gate points (24 more tied under the 1 dB floor, of 134 comparable) and best-LTE at 9 of 105 decided (29 more tied under the 1 dB floor) (LTE median −93.7 dBm, 78 of 134 healthy). It reports 5G on 59% of its on-airport tiles (panel-sensitive). Inside the footprints its healthy share of 5G-reporting tiles runs 30% to 49%.

Tied at the venue gate-area median; leads inside 5 of 5 terminal footprints (building-masked); best-5G at the most gate points in Terminal 5 and Terminal 7 (gate-area counts). Venue gate-area median −99.3 dBm, healthy at 33 of 134 gate points (25%), with a 5G SINR median of +10.6 dB (71 of 134 healthy). It is best-5G at 44 of 110 decided gate points (24 more tied under the 1 dB floor, of 134 comparable) and best-LTE at 5 of 105 decided (29 more tied under the 1 dB floor) (LTE median −97.3 dBm, 35 of 134 healthy). It reports 5G on 78% of its on-airport tiles (panel-sensitive). Inside the footprints its healthy share of 5G-reporting tiles runs 62% to 82%. Its LTE layer is the weakest of the three.

Tied at the venue gate-area median; leads inside 0 of 5 terminal footprints (building-masked); best-5G at the most gate points in no terminal (gate-area counts). Venue gate-area median −100.2 dBm, healthy at 9 of 134 gate points (7%), with a 5G SINR median of +8.0 dB (38 of 134 healthy). It is best-5G at 12 of 110 decided gate points (24 more tied under the 1 dB floor, of 134 comparable) and best-LTE at 91 of 105 decided (29 more tied under the 1 dB floor) (LTE median −90.8 dBm, 102 of 134 healthy). It reports 5G on 45% of its on-airport tiles (panel-sensitive). Inside the footprints its healthy share of 5G-reporting tiles runs 10% to 60%. Its LTE layer is the venue LTE network of record.

JFK is operated by the Port Authority of New York and New Jersey. The record we hold describes a large hub with five operating terminals, Boingo Wireless as the neutral-host DAS and Wi-Fi provider of record, and a terminal program that replaces two of the measured buildings. Measured facts in this report stand on the tile export alone; public-record facts are tagged inline and never originate a measurement.

Photo · Terminal 4 apron and gates, September 2024Wikimedia Commons

Photo: Oleg Yunakov, CC0 (2024-09-15), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Fig. 1 · Venue map · aerodrome, OSM terminal buildings, 134 gate points, 50,923 unique RF tilesSource · Ookla

Aerodrome = OSM way 158042008 plus 200 m clip. NTO is New Terminal One, under construction. North up; scale bar in meters. The east strip beyond the export rectangle carries no tiles. Source: Ookla Cell Analytics tiles on OpenStreetMap geometry. Window 2025-09-01 to 2026-08-31.

FAA airport diagram · reproduced for orientationWikimedia Commons

Federal Aviation Administration, Public domain (2025-01-01), via Wikimedia Commons. It is the FAA's drawing, not this report's geometry; runway and taxiway labels come from the FAA.

Carriers are printed strongest at the gate points first (Verizon, T-Mobile, AT&T), asserted against the data at build time. Each gate-area value is the arithmetic mean of tiles within 80 m of the gate point, which straddles the apron; the venue value is the median across the 134 gate points. Every published gate-carrier pair has at least 30 tiles. Per-gate best-carrier calls use the same 1 dB floor: a carrier is named only when its top-two margin reaches 1 dB, and smaller margins print as a tie with the margin (24 tied on 5G NR, 29 on LTE, of 134 comparable gate points). The adjacent margins (1.0 and 0.8 dB) are under the 1 dB noise floor, so the venue order is a presentation order, not a ranking.

Healthy = gate-area mean ≥ −95 dBm; n = 134 gate points per carrier. SINR healthy (≥ 10 dB): Verizon 45, T-Mobile 71, AT&T 38 of 134. LTE healthy gate points: AT&T 102, Verizon 78, T-Mobile 35 of 134. Full per-gate tables ship in the PDF appendices.

Each 5G median is conditional on the tile reporting a 5G reading; the footprint share is panel-sensitive (a no-reading tile may be a panel gap on that carrier). n counts tiles, not scans.

AT&T's two layers diverge more than the other carriers'. Its LTE median at the gate points is −90.8 dBm (102 of 134 healthy) against a 5G median of −100.2 dBm, and 53% of its on-airport tiles carry only an LTE reading. Inside the footprints its 5G footprint share is T1 70%, T4 67%, T5 63%, T7 39%, T8 60% against T-Mobile's 98%, 96%, 96%, 93%, 94%, so where AT&T reports 5G it is a smaller, strong-signal subset; its healthy share of reporting tiles is 10% to 60%. Its weakest footprint is Terminal 1 (10% healthy of n=226 reporting tiles, median −109.0 dBm), a building that the New Terminal One program replaces cited source.

Photo · Terminal 8 exterior, June 2024Wikimedia Commons

Photo: 颐园居, CC BY 4.0 (2024-06-02), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Fig. 2 · Venue medians across 134 gate points · 5G RSRP and SINRSource · Ookla

RSRP on a −110 to −85 dBm axis (SINR on 0 to 15 dB). Locked ladder: healthy ≥ −95 · at-risk −95 to −110 · critical < −110 dBm. Window 2025-09-01 to 2026-08-31.

Fig. 3 · SINR distribution per carrier at the gate pointsSource · Ookla

SINR healthy threshold ≥ 10 dB. T-Mobile is the SINR leader at the gate-area median (+10.6 dB, 71 of 134 healthy). Window 2025-09-01 to 2026-08-31.

Fig. 4 · 5G NR RSRP by tile · terminal area · locked coverage classSource · Ookla

5G NR RSRP by tile, terminal area, locked coverage class; grey = no 5G reading. Each tile is about 8 m by 11 m; n is the tile count per panel, printed on the figure. Gate labels are off for legibility. Source: Ookla Cell Analytics. Window 2025-09-01 to 2026-08-31.

Fig. 5 · LTE RSRP by tile · terminal area · locked coverage classSource · Ookla

LTE RSRP by tile, terminal area, locked coverage class; grey = no LTE reading. Same extent and tile geometry as Fig. 4; gate labels are off for legibility. This is the LTE layer that carries AT&T's traffic where its 5G footprint is thin. Source: Ookla Cell Analytics. Window 2025-09-01 to 2026-08-31.

Two bases. Building-masked: tiles whose centroid lies inside the pinned OpenStreetMap terminal polygon; the healthy share of 5G-reporting tiles inside it decides the terminal leader (n ≥ 30 reporting tiles per carrier; margins under 5 points are ties; the carrier's 5G footprint share inside the polygon is printed beside it). Gate-area: the mean of tiles within 80 m of each gate point, which straddles the apron (terminal gate counts T1 16, T4 47, T5 29, T7 10, T8 32; mean interior share T1 24%, T4 22%, T5 33%, T7 41%, T8 30%). Gate-area tallies are counts with n visible and are never rankings. New Terminal One holds no gates and no tenant points in the window and appears only as a count-only footprint.

Building-masked basis. Per carrier: 5G-reporting tiles inside the polygon (n) of all tiles inside · footprint share (fp) · healthy share of reporting tiles · median 5G RSRP of reporting tiles. Leader margin in points of healthy share. n counts tiles, not scans.

Leader list (printed once; referenced everywhere else). Decisive: Terminal 1 T-Mobile by 13 points; Terminal 4 T-Mobile by 30 points; Terminal 5 T-Mobile by 10 points; Terminal 7 T-Mobile by 34 points; Terminal 8 T-Mobile by 42 points. Ties: none. Cross-check: the leader on the panel-sensitive present-and-healthy share of all tiles inside the polygon agrees with the conditional leader in every ranked terminal. Conditional-median leader by terminal: T1 T-Mobile, T4 T-Mobile, T5 T-Mobile, T7 T-Mobile, T8 T-Mobile. Tally of decisive leads: T-Mobile 5.

Fig. 6 · Building-masked healthy share of 5G-reporting tiles per terminal and carrierSource · Ookla

The terminal-leader basis. fp = each carrier's 5G footprint share inside the polygon. Window 2025-09-01 to 2026-08-31.

Fig. 7 · Building-masked median 5G RSRP of reporting tiles per terminal and carrierSource · Ookla

Fig. 8 · Building-masked map · 5G NR class for reporting tiles inside the OSM terminal polygonsSource · Ookla

5G-reporting tiles whose centroid falls inside a pinned OSM terminal polygon (T1, NTO, T4, T5, T7, T8). This is our OSM building-polygon mask, not Ookla's Indoor layer. Source: Ookla Cell Analytics on OpenStreetMap footprints. Window 2025-09-01 to 2026-08-31.

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 them. It is the only lens in this export that reads the buildings rather than the apron, which is why it decides the terminal leaders and the section 13 criteria.

Critical-band gate points (below −110 dBm) by carrier: Verizon 0, T-Mobile 0, AT&T 12 of 134 (counts). On the pool basis (nearest gate or tenant point within 80 m; never the leader basis), the critical share peaks for Verizon in Terminal 5 at 15% of reporting tiles (n=1,057); T-Mobile in Terminal 8 at 11% of reporting tiles (n=2,039); AT&T in Terminal 1 at 37% of reporting tiles (n=457).

Fig. 9 · Coverage-class distribution by terminal and carrier · 80 m gate-and-concession poolSource · Ookla

V = Verizon, T = T-Mobile, A = AT&T. Conditional on a reading; not the leader basis. Window 2025-09-01 to 2026-08-31.

Apron caveat first. Each gate point's value is the mean of tiles within 80 m; the share of those tiles whose centroid lies inside a terminal polygon (the interior share) averages T1 24% (median 28%), T4 22% (median 23%), T5 33% (median 30%), T7 41% (median 42%), T8 30% (median 31%). 7 of 134 gate points have an interior share at or above 50% (T1 0, T4 0, T5 4, T7 2, T8 1). Gate-area readings therefore describe the pier edge and the stand more than the concourse.

By terminal, all-weak gate points: T1 2 of 16, T4 43 of 47, T5 14 of 29, T7 4 of 10, T8 12 of 32.

Interior re-cut (building-masked; a carrier is weak inside a footprint when its healthy share of 5G-reporting tiles is below 50%): Terminal 1 2 of 3 weak (Verizon, AT&T); Terminal 4 2 of 3 weak (Verizon, AT&T); Terminal 5 1 of 3 weak (Verizon); Terminal 7 2 of 3 weak (Verizon, AT&T); Terminal 8 2 of 3 weak (Verizon, AT&T). No footprint is weak for all three carriers on the interior re-cut, while 75 of 134 gate points are weak for all three on the gate-area basis. Read together: the apron-straddling tile sets are weak for everyone, and inside the buildings the weakness is carrier-specific.

Fig. 10 · Carriers below the healthy line per gate point, and the building-interior re-cutSource · Ookla

Left: gate-area counts (mean interior share on the axis). Right: building-masked healthy share of 5G-reporting tiles inside each footprint per carrier against the interior weak line. Window 2025-09-01 to 2026-08-31.

Gate-area basis. The locked −95 dBm line is the Imagine Wireless Venue Network Intelligence ladder; no client-specified design target was provided, and the building-masked basis in section 4 uses the locked line.

Terminal 4 hosts Delta plus international carriers (Emirates, Virgin Atlantic, Etihad, KLM, Singapore, LATAM); operated by JFKIAT (Tier B) and holds 47 of the 134 gate points. Building-masked: T-Mobile leads by 30 points; healthy shares of reporting tiles Verizon 36% (n=558, footprint 63%), T-Mobile 69% (n=885, footprint 96%), AT&T 39% (n=530, footprint 67%); interior re-cut 2 of 3 carriers weak (Verizon, AT&T). Gate-area (counts; mean interior share 22%): medians Verizon −100.6 dBm (0 of 47 healthy), T-Mobile −102.0 dBm (2 of 47 healthy), AT&T −105.5 dBm (4 of 47 healthy); best-5G Verizon 24, AT&T 9, T-Mobile 6, 8 tied under the floor; all-weak 43 of 47.

Fig. 11 · Per-gate 5G RSRP · 47 Terminal 4 gate points · gate-area basisSource · Ookla

OpenStreetMap inventory after same-ref merge. Each bar is a mean over at least 30 tiles within 80 m of the gate point. Window 2025-09-01 to 2026-08-31.

Terminal 8 hosts American plus oneworld partners (British Airways, Finnair, Iberia, Japan Airlines, Qatar, Cathay Pacific) and Alaska Airlines (Tier B; Alaska Airlines moved from Terminal 7 to Terminal 8 in October 2025, cited, not held) and holds 32 gate points. Building-masked: T-Mobile by 42 points; healthy shares Verizon 40% (n=311, footprint 62%), T-Mobile 82% (n=521, footprint 94%), AT&T 35% (n=311, footprint 60%); interior re-cut 2 of 3 weak. Gate-area (counts; mean interior share 30%): medians Verizon −96.7 (10 of 32 healthy), T-Mobile −97.0 (13 of 32 healthy), AT&T −99.1 (0 of 32 healthy) dBm; best-5G Verizon 14, T-Mobile 13, 5 tied under the floor.

Terminal 5 hosts JetBlue domestic and daytime international arrivals (Tier B) and holds 29 gate points. Building-masked: T-Mobile by 10 points; healthy shares Verizon 34% (n=385, footprint 63%), T-Mobile 70% (n=618, footprint 96%), AT&T 60% (n=334, footprint 63%); interior re-cut 1 of 3 weak. Gate-area (counts; mean interior share 33%): medians Verizon −97.7 (0 of 29 healthy), T-Mobile −95.2 (13 of 29 healthy), AT&T −97.4 (4 of 29 healthy) dBm; best-5G T-Mobile 18, AT&T 2, 9 tied under the floor.

Fig. 12 · Per-gate 5G RSRP · 61 Terminal 5 and Terminal 8 gate points · gate-area basisSource · Ookla

Each bar is a mean over at least 30 tiles within 80 m of the gate point. Window 2025-09-01 to 2026-08-31.

Photo · JetBlue aircraft at a JFK gate, August 2023Wikimedia Commons

Photo: Tim, CC0 (2023-08-08), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Photo · Terminal 8 concourse interiorWikimedia Commons

Photo: 颐园居, CC BY-SA 4.0 (2024-06-02), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Terminal 1 hosts legacy Terminal 1 international carriers (moving to New Terminal One or Terminal 6 after the window) (Tier B) and holds 16 gate points. Building-masked: T-Mobile by 13 points; healthy shares Verizon 49% (n=192, footprint 63%), T-Mobile 62% (n=330, footprint 98%), AT&T 10% (n=226, footprint 70%); interior re-cut 2 of 3 weak. Gate-area (counts, below the floor; mean interior share 24%): best-5G Verizon 16; healthy Verizon 14 of 16, T-Mobile 0 of 16, AT&T 0 of 16; all-weak 2 of 16. Verizon is best-5G at 16 of 16 Terminal 1 gate points and healthy at 14 of 16 on the gate-area basis (counts, below the 30-gate floor; mean interior share 24%); the gate points sit on the piers and the footprint covers the headhouse, so the two bases describe different parts of the building.

Photo · New Terminal One under construction, August 2025Wikimedia Commons

Photo: Charles from Port Chester, New York, CC0 (2025-08-02), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Photo · Terminal 7 departures entrance, February 2025Wikimedia Commons

Photo: Quintin Soloviev, CC BY 4.0 (2025-02-06), via Wikimedia Commons. Illustrative (Tier B), resized only; the photograph is not a measurement.

Terminal 7 hosts carriers transitioning to Terminal 6 from late 2026; Alaska left for T8 in October 2025 (Tier B) and holds 10 gate points. Building-masked: T-Mobile by 34 points; healthy shares Verizon 30% (n=92, footprint 46%), T-Mobile 77% (n=200, footprint 93%), AT&T 43% (n=75, footprint 39%); interior re-cut 2 of 3 weak. Gate-area (counts, below the floor; mean interior share 41%): best-5G T-Mobile 7, AT&T 1, 2 tied under the floor; healthy Verizon 0 of 10, T-Mobile 5 of 10, AT&T 1 of 10; all-weak 4 of 10.

Fig. 13 · Per-gate 5G RSRP · 26 Terminal 1 and Terminal 7 gate points · counts onlySource · Ookla

Gate-area basis; OpenStreetMap inventory; counts only, below the 30-gate floor. Window 2025-09-01 to 2026-08-31.

Both terminals are on the program's replacement list. New Terminal One (NTO; $9.5 billion, 23 gates) rises on the Terminal 1 site and the former Terminals 2 and 3 site and is not open in the measurement window; the first phase is now expected in early 2027. Terminal 6 ($4.2 billion, 10 gates) has its first gates expected in late 2026, replacing the former Terminal 6 and eventually Terminal 7. Terminal 2 has been demolished and has no footprint in OpenStreetMap. All cited source. In-building investment in Terminal 1 and Terminal 7 should therefore be scoped to their remaining service life, and the new buildings are where the section 13 criteria should be written into the design.

This proxy compares each carrier's median 5G RSRP over tiles near terminal-category points (retail, food, lounge, admin, services; 299 points) with tiles near periphery-category points (transport admin, lodging, other; 115 points). Terminal categories cover the concessions and services inside the terminal footprints; periphery categories cover the parking lots, the AirTrain stations, the hotel, the cargo area, and roadside businesses. A positive periphery-minus- terminal delta is consistent with macro spillover; a negative delta is consistent with an in-building or close-in source; a flat delta is ambiguous.

Distance to macro sites confounds all three, and the Site Finder layer is queued to resolve it. The building-masked medians in section 4 and the reconciled reading in section 9 are the sharper in-building evidence.

Fig. 14 · Median 5G RSRP · terminal vs periphery tenant categoriesSource · Ookla

Proxy: category-based masking on 414 mapped points (Google Places + OpenStreetMap, pinned 2026-09-15); the building-masked view (section 4) is the sharper in-building lens. Window 2025-09-01 to 2026-08-31.

Boingo Wireless is the Port Authority's neutral-host DAS and Wi-Fi provider at JFK (Boingo press release, October 14, 2025, held; PANYNJ board authorization, June 26, 2025, Imagine knowledge graph; Airport DAS Master, July 2026, Imagine knowledge graph). Tier B · held

The agreement was authorized on June 26, 2025 under the 2024 RFP, announced on October 14, 2025 as a long-term agreement to deliver next-generation cellular and free public Wi-Fi across Port Authority properties including JFK, runs 2026 to 2038, and replaces the 1999 New York Telecom Partners agreement under which Boingo already operated at JFK; Boingo also provides Passpoint Wi-Fi at JFK. Boingo Wireless, through its subsidiary New York Telecom Partners, has provided neutral-host distributed antenna system (DAS) and Wi-Fi services at JFK, LaGuardia, and Newark since the 1999 agreement (Boingo press release, August 11, 2014).

The sources we hold do not map which terminal buildings ride on Boingo's network versus terminal-operator systems, nor which carriers are on the DAS in each building; ask Boingo for that map.

On the Boingo-served venue, the tiles show that the building-masked readings favor T-Mobile inside every ranked footprint and Verizon at the Terminal 1 gate points, both consistent with a close-in or in-building source (an inference about the radio source; the report does not assign carriers to systems), while the apron-straddling gate-area medians sit below the healthy line for every carrier.

No Citizens Broadband Radio Service (CBRS) or public-safety private-network deployment is recorded for JFK in the sources this report holds, and the public Ookla layer does not characterize private networks. The two JFK-specific questions are whether a CBRS layer is scoped alongside the neutral-host upgrade for the New Terminal One and Terminal 6 construction phases, and whether the federal inspection and baggage workflows in Terminals 4 and 8 ride on the public carriers or on a dedicated layer; both are design questions for the Port Authority and the terminal operators, not measurements.

The 5G footprint is the share of a carrier's on-airport tiles that report a 5G NR (New Radio) reading, and it is panel-sensitive: a no-reading tile may be a panel gap on that carrier. T-Mobile 78% (33,821 of 43,109 tiles), Verizon 59% (17,066 of 28,746), AT&T 45% (10,650 of 23,509). 53% of AT&T's on-airport tiles carry only an LTE reading, which is why its LTE median at the gate points (−90.8 dBm) matters more than it does for the other two carriers.

Fig. 15 · 5G footprint · share of tiles with a 5G NR readingSource · Ookla

Footprint counts tiles with a non-null 5G reading. Panel-sensitive. Window 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. Verizon mean 6.5%, median 0% (tiles above zero: 27% of n=28,109) · T-Mobile mean 3.2%, median 0% (tiles above zero: 22% of n=42,526) · AT&T mean 15.4%, median 0% (tiles above zero: 40% of n=23,018).

The LTE metric does not see 5G low band. On the 5G side: Verizon NR850 on 759 tiles, 13.8% of its 5,506 band-reporting NR tiles. T-Mobile NR600 on 4,273 tiles, 15.5% of its 27,561 band-reporting NR tiles. AT&T NR850 on 95 tiles, 23.2% of its 409 band-reporting NR tiles. LTE700PS, FirstNet Band 14, appears on 509 AT&T LTE tiles. Cell, PCI, and band fields populate only where Ookla resolved a serving cell, so band n is smaller than RSRP n; AT&T's NR band mix rests on the 409 tiles where an NR serving cell was resolved and describes a subset.

Fig. 16 · LTE percent-low-band distribution per carrierSource · Ookla

Higher means more attachment to sub-1 GHz coverage bands. Macro-site geometry is queued for the next revision. Window 2025-09-01 to 2026-08-31.

Tile medians per carrier. CQI (0 to 15) is the closest throughput proxy in an export that carries no throughput fields. Optimization priority is Ookla's vendor-defined index. Volatility = share of tiles where the strongest observed cell differs from the most-frequent serving cell. Availability is the mean over reporting tiles only.

Inference: an NR serving cell that resolves on 4% of AT&T's 5G tiles is consistent with non-standalone (NSA) anchoring, where the handset reports the LTE anchor as its serving cell; it would also fit AT&T's 53% LTE-only share. The export cannot confirm it. AT&T's 5G volatility and PCI diversity therefore describe a 409-tile subset and are not comparable with the other carriers' bars.

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. Terminal medians and 95th percentiles (p95) are on the gate-and-concession pool.

Density-weighted ranking = weakest-carrier below-healthy share inside the footprint (building-masked, conditional) × median density. Wi-Fi share of observed data volume = Σ Wi-Fi MB ÷ Σ (Wi-Fi + mobile) MB on the All Networks panel.

Terminal 1 leads the density-weighted ranking (AT&T below healthy on 90% of its reporting tiles inside, median density 6). The weakest carrier inside the footprint is AT&T in T1, Verizon in T4, Verizon in T5, Verizon in T7, AT&T in T8. The on-airport median is 3 devices per tile across 50,663 usage tiles. Usage volume is observed sample volume from the crowdsource panel. It is not total traffic.

Fig. 17 · Worst-carrier 5G class per tile and unique-device density per tileSource · Ookla

Left: worst-carrier 5G NR class per tile where all three carriers report. Right: User_Density per usage tile, unique devices over the window, deduplicated, log scale. Amber dominates the apron edges on the worst-carrier panel, and the density panel shows the terminal cores and the AirTrain corridor carrying the highest unique-device counts. Source: Ookla Cell Analytics. Window 2025-09-01 to 2026-08-31.

Fig. 18 · Unique-device density by terminal · density-weighted weak-share rankingSource · Ookla

User_Density = unique devices over the window, deduplicated. Density on the gate-and-concession pool; ranking on the building-masked conditional basis. Window 2025-09-01 to 2026-08-31.

Venue-wide, 56.2% of observed data volume moves over Wi-Fi (n=50,663 usage tiles). Terminal shares run from 50% in Terminal 5 to 61% in Terminal 1. Wi-Fi usage is observed on 57% of the 40,920 on-airport usage tiles that report the Wi-Fi field.

Scope decomposition: tiles assigned to a terminal pool blend to 57.5%; the remaining on-airport tiles (airfield, roadways, cargo, remote lots) run 54.2% Wi-Fi, pulling the venue figure to 56.2%. 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 weakest carrier is weakest inside the building and Wi-Fi presence is high.

Fig. 19 · Observed data volume by terminal and bearerSource · Ookla

Fig. 20 · Wi-Fi share of observed data volume by terminalSource · Ookla

Per gate point, the atlas names the weakest carrier and the spread from the strongest carrier's mean to the weakest. Worst-carrier tally across all 134 ranked gate points: AT&T 90, Verizon 33, T-Mobile 11. Every carrier's per-gate n is at least 30 tiles at all 134 gate points, so none are excluded. The top 20 gate points are where a passenger's experience varies most by carrier on the gate-area basis (spreads of 9 to 22 dB). The largest spreads sit in Terminal 1 (14 of the top 20); their interior shares average 25%, so they are pier-edge readings. They serve two purposes: worst-case examples for the stakeholder narrative, and the gate-point validation list for section 13.

Fig. 21 · Carrier-asymmetry atlas · top 20 gate points by spreadSource · Ookla

Spread = best minus worst carrier mean 5G RSRP at the gate point, among the 134 gate points where every carrier's per-gate n is at least 30 tiles (gate-area basis; interior share on the label). Window 2025-09-01 to 2026-08-31.

Criteria reference the building-masked basis (section 4) and the monitoring table below.

1. Ask Boingo for the per-terminal coverage map

   The sources we hold do not map which terminal buildings ride on Boingo's network versus terminal-operator systems, nor which carriers are on the DAS in each building; ask Boingo for that map, with the phases and go-live dates of the 2025 agreement; every design question below depends on it.

   Owner · Port Authority (with Boingo)

2. Confirm each carrier's participation on the Boingo system per terminal

   The building-masked readings (section 9) favor T-Mobile inside every ranked footprint and Verizon at the Terminal 1 gate points, consistent with a close-in or in-building source; Boingo and the carriers can confirm which carriers are live on the DAS in each building and since when.

   Owner · Port Authority with Boingo and the carriers

3. Write the monitoring criteria into the New Terminal One and Terminal 6 designs

   For every hosted carrier, a healthy share of 5G-reporting tiles at or above 90% inside the footprint on a like-for-like re-measure (this window's best inside any footprint is 82%), and a healthy 5G median at every interior gate point.

   Owner · Port Authority, terminal developers

4. Scope Terminal 1 and Terminal 7 in-building work to their remaining service life

   Inside the Terminal 1 footprint AT&T is healthy on 10% of its 226 reporting tiles (median −109.0 dBm) and New Terminal One replaces the building (cited source); a like-for-like fix there should be a bridge, not a build.

   Owner · Port Authority, terminal operators

5. Treat Terminal 4 as the priority existing building on the interior re-cut

   2 of 3 carriers weak inside the footprint (Verizon, AT&T) on the largest interior tile set (1,973 reporting tiles across carriers); Terminal 1, Terminal 7, Terminal 8 tie on the weak-carrier count with smaller interior sets. Its operator (JFKIAT, Tier B) should be the first design partner. The gate-area count (43 of 47 gate points weak for all three) is an apron-edge reading and is not the justification.

   Owner · JFKIAT, Port Authority

6. Treat Wi-Fi as a coexisting capacity layer in the design

   Wi-Fi share of observed data volume 56%, section 11. Obtain controller telemetry to size it, and decide the CBRS and private-wireless question alongside the neutral-host upgrade (section 9), not after it.

   Owner · Port Authority, Boingo, terminal operators

1. Re-measure over a like-for-like window after any in-building change goes live

   The same 12-month length, or a matched shorter window that is disclosed. Crowdsourced tiles serve as the before-and-after comparison; acceptance itself is per-carrier commissioning against the same thresholds.

   Owner · Imagine Wireless

2. Pull the queued layers before the next issue

   Site Finder centroids, the knowledge-graph contracts and artifact rows, and the Ookla Indoor layer captures. Each upgrades a named claim from proxy to measurement.

   Owner · Imagine Wireless

3. Reconcile the OpenStreetMap gate inventory against each terminal operator's gate plan

   So gate-named criteria carry the operator's labels.

   Owner · Imagine Wireless with terminal operators

These are monitoring criteria for the before-and-after comparison, not the acceptance test for any system. Acceptance itself is per-carrier commissioning (walk test and scanner survey) against the same thresholds. Criteria are stated on the building-masked conditional set (healthy share of 5G-reporting tiles inside each footprint) and on the 7 gate points whose 80 m tile set is at least 50% inside a footprint; the panel-sensitive present-and-healthy share is not a criterion. Gate refs come from the OpenStreetMap inventory after the same-ref merge and must be reconciled against each terminal operator's gate plan before adoption.

Re-test for every row: like-for-like window after go-live, same instrument (Ookla Cell Analytics, identical thresholds). New Terminal One and Terminal 6 enter the footprint set when their OpenStreetMap polygons and gates exist.

New Terminal One replaces Terminal 1 cited source. The same criteria apply for as long as the building operates, and are listed separately so they are not written into a permanent design.

Re-test: like-for-like window while the building operates.