Confidential client deliverable · Enter the passcode you were given
ONT · Deep Edition
An independent, measurement-grounded read on the airport's in-building cellular and Wi-Fi: characterized gate by gate and tenant by tenant, and framed to the open Private 5G procurement Imagine knowledge graph.
ONT runs three public carriers across two terminals and the International Arrivals building, with one registry-recorded in-building system: a Verizon single-carrier DAS, active since 2022 Imagine knowledge graph · confidence 0.80. The measurement reads exactly like that record predicts. One carrier's DAS is doing exactly what a DAS is for — and the other two are outside it.
The strongest signal and the cleanest quality in the building: median 5G strength −82 dBm (n=263 tiles) and median quality +23 dB (n=261) across the terminal footprints, best-serving 5G at all 27 of 27 gates, and 0 of 27 gates below the healthy line. Its signal is stronger inside the building than on the apron (−82 indoor vs −88 outdoor, +6 dB indoor gain) — the fingerprint of a working in-building system, and consistent with the registry-recorded DAS.
The widest 5G footprint of the three — 5G present on 98% of its terminal-footprint tiles — but it rides macro spillover at the edge of healthy: median 5G strength −95 dBm (n=331), exactly on the healthy line, with median quality +5 dB and 21 of 27 gates below healthy. The signal is flat across the building skin (−96 apron to −95 indoor), which is what outdoor coverage reaching in looks like: broad and usable, with thin quality margin.
Weak on both sides of the building skin: median 5G strength −103 dBm indoors (n=203) and −105 dBm on the apron (n=525). A thin local macro grid plus no in-building path makes AT&T the venue's largest addressable gap — below healthy 5G at all 27 gates, weakest 4G at all 27, and concentrated in Terminal 4, where 88.7% of its footprint 4G tiles read below healthy (n=141).
The carrier order above is computed, not chosen: descending median 5G strength over the terminal footprints. Verizon's registry-recorded DAS explains its inversion of the outdoor-to-indoor norm; T-Mobile's flat skin-crossing matches macro spillover; AT&T is weak outdoors too, so its problem is not donor isolation alone. The deployment record explains the measurement; the measurement stands without the record.
No gate at ONT is weak for all three carriers at once — of 27 gates, 21 are weak for exactly two carriers and 6 for exactly one Derived. There is no building-wide propagation failure to fix; the weakness is carrier-specific. And 4G service availability is effectively universal for all three carriers, so passengers run thin, not dark.
A Private 5G procurement at ONT is at RFP stage Imagine knowledge graph. That procurement is the venue's natural decision moment for the public-carrier question too: whether the answer is a neutral-host extension, a multi-carrier DAS scope, or carrier-funded remediation, the baseline in this report is what any of those get written against. Section 14 turns it into acceptance criteria.
Ontario International is a Medium Hub with 3,494,554 annual enplanements Imagine knowledge graph, operated by the Ontario International Airport Authority. Here is the wireless environment as the data and the airport records describe it today. Knowledge-graph facts carry an inline tag; measurement figures carry the data window and their n.
| Element | Value | Notes |
|---|---|---|
| Airport | Ontario International · ONT / KONT | Ontario International Airport Authority (OIAA) · FAA Medium Hub · 3,494,554 enplanements (latest FAA calendar-year figure; registry snapshot 2026-07-02) Imagine knowledge graph |
| Terminals · gates | 2 terminals + the International Arrivals building · 27 gates | Terminal 2 (13 gates) · Terminal 4 (14 gates) · International Arrivals (no gates in scope) |
| Tenants | 54 | Terminal-attributed tenants: food, retail, services, transport/admin |
| Carriers measured | 3 | Verizon · T-Mobile · AT&T · 29,205 carrier tile-aggregates + 16,267 usage tiles |
| Registry-recorded in-building cellular | Verizon single-carrier DAS | Active since 2022 Imagine knowledge graph · confidence 0.80. The measurement is consistent with this record (Section 3), and the measured asymmetry stands independently of it |
| Open procurement | Private 5G · RFP stage | Imagine knowledge graph · verify against the live procurement record before external use. The decision moment this report is framed to |
| Bin geometry | 0.0001° ≈ 11 × 9 m | Coordinate-rounded bins; measured pitch. No Z-axis; jet bridges and back-of-house are out of scope until a walk-test layer is added |
| Comparable measured venues | SEA · MIA · LAX | Same platform, same threshold ladder, same method standard (Section 16) |
Every figure traces to the Ookla Cell Analytics tile dataset over the twelve-month window 2025-06-01 to 2026-05-31, summarized onto gate envelopes and terminal footprints from a pinned OpenStreetMap snapshot and tenant points from Google Places. Headline figures are medians; means appear only where labeled. No chart or comparative claim in this report rests on fewer than 30 tiles or points; smaller cells are reported as counts with their n visible and are never ranked.
3.0Two things decide mobile experience: how strong the signal is, and how clean it is. At most venues the two views disagree and split the carriers between them. At ONT they agree completely: Verizon wins both, and AT&T loses both. That is the whole story of this venue in one figure.
Medians over terminal-footprint tiles. n (strength / quality): Verizon 263 / 261 · T-Mobile 331 / 331 · AT&T 203 / 194. Thresholds: strength healthy at −95 dBm or better; quality healthy at +10 dB or better. Hover a carrier for detail.
Per-gate per-carrier mean 5G strength vs −95 dBm. Every per-gate carrier base is ≥ 30 tiles on the all-tile basis (per-gate n in Appendix A); 5G-reporting bases can be smaller, and the two sub-floor cells (AT&T at gates 401/402) are marked indicative in Appendix A and excluded from ranked views. The min-n floor governs tiles per cell — per-gate diagnostic tables are published with per-row n.
Verizon holds the strongest and cleanest 5G in the building at once, and its 4G median of −80 dBm (n=299) leads as well. The decisive evidence is the direction of the skin-crossing: −88 dBm on the apron and curb (n=896) rising to −82 dBm inside the footprint (n=263). Outdoor networks lose strength crossing a building skin; a gain in that direction is consistent with a dedicated in-building system, and rarely produced without one. That inversion is measured independently of the deployment record — and the record, a single-carrier Verizon DAS active since 2022 Imagine knowledge graph, is exactly what would produce it. The as-built documentation task in Recommendation 2 is what converts this inference into fact.
94.0% of Verizon's footprint 4G tiles read healthy, 6.0% at-risk, 0.0% critical (n=299). Verizon is the venue's proof that the building is coverable: the propagation environment supports healthy service everywhere a system is actually built to deliver it.
T-Mobile's 5G reaches almost everywhere — present on 98% of its terminal-footprint tiles, the widest of the three — but at the edge of what the ladder calls healthy. The median sits exactly on the −95 dBm line, quality holds at +5 dB, and the signal barely changes crossing the building skin (−96 apron, n=1,306, to −95 indoor, n=331). That flat profile is macro coverage reaching in: broad and usable, with no in-building source adding headroom. On 4G the same picture: 62.4% of footprint tiles healthy, 36.2% at-risk, 1.4% critical (n=282).
Two disclosures govern this finding. First, threshold sensitivity: the 21-of-27 below-healthy count rides on gate medians clustered at the −95 dBm line itself — shift the ladder 1 dB and the count moves sharply — so the durable signal is the +5 dB median SINR margin, not the gate count. Second, this export carries no throughput or load data; reading thin margin as capacity risk is an engineering inference, labeled as such, not a load measurement. It is why the density view in Section 6 matters more for T-Mobile than its medians suggest.
AT&T is not a donor-isolation story, because it is weak outdoors too: −105 dBm median on the apron (n=525) against −103 dBm inside (n=203). A thin local macro grid plus no in-building path leaves nothing for the building to keep out. On 4G, only 15.9% of footprint tiles read healthy, 72.6% at-risk, and 11.5% critical (n=314) — and the split worsens in Terminal 4: 11.3% healthy / 73.0% at-risk / 15.6% critical (n=141) against Terminal 2's 22.0 / 69.5 / 8.5% (n=141).
AT&T is weakest 4G at every one of 27 gates and below healthy 5G at every one, with the worst of it concentrated in Terminal 4 — the terminal where the weakest worst-carrier share, terminal-grade device concentration, and the venue's lowest Wi-Fi offload stack together (Sections 6 and 7). Because the cause is structural — thin grid, no indoor path — remediation is buildable: macro augmentation, an indoor path, or both. Section 13 makes engaging AT&T with this evidence a named recommendation.
The share of a carrier's 4G samples attaching below 1 GHz separates capacity-grade attachment from coverage-grade survival — samples are scan counts, not data volume. These are labeled, cleaned means over each carrier's tiles — per-tile medians are 0% for every carrier, so the metric is heavily skewed and read only directionally Derived.
AT&T's reliance on coverage-grade spectrum is roughly four times T-Mobile's, corroborating the thin-grid finding in 3.3 from an independent angle. Export artifact, disclosed and excluded: 51 of 28,775 tiles report a low-band share above 100% (max 400) and are removed from these cleaned means (data-layer means including them: 19.4 / 8.2 / 4.9, pinned in the audit ledger).
Signal-quality corroborators point the same way as the headline medians: 5G RSRQ medians run Verizon −11 dB, T-Mobile −12, AT&T −14, and 4G channel-quality (CQI) medians run 9 / 9 / 8. The separation is weak — consistent with the main findings, not independent confirmation of them — and no claim in this report leans on these fields alone.
4.1ONT's zones are the two terminal footprints, their apron and curb surroundings, and the International Arrivals building. The locked ladder — healthy ≥ −95 dBm, at-risk −95 to −110, critical below −110 — applied to footprint 4G tiles separates the carriers cleanly.
Share of footprint tiles reporting 4G strength per class. Ladder: healthy ≥ −95 dBm · at-risk −95 to −110 · critical < −110. The AT&T rows split its footprint by terminal; all bases n ≥ 141. Hover a segment for counts.
Comparing each carrier's 5G median on the apron and curb against the terminal footprint reads the building skin directly Derived. Only one carrier gains strength crossing inward.
Verizon inverts the outdoor-over-indoor norm (+6 dB indoor gain) — consistent with a dedicated in-building system, and rarely produced without one; the as-built documentation (Recommendation 2) converts the inference to fact. T-Mobile is flat (+1 dB); AT&T is weak on both sides (+2 dB on a −105 base). n per bar shown on hover.
The stronger terminal for AT&T (22.0% healthy 4G, n=141) and the venue's Wi-Fi leader at 54.3% offload. Verizon healthy throughout; T-Mobile at the edge of healthy.
Where everything stacks. Device concentration equal to T2's (median 21 unique devices per bin, n=157 tiles), the venue's lowest Wi-Fi offload (34.1%), and AT&T's worst split: 11.3% healthy / 15.6% critical 4G (n=141). The widest ranked carrier-asymmetry gates are all here.
The weakest-measured building — and the thinnest-measured. Per-carrier 5G bases fall below the 30-tile floor for two of three carriers (Verizon n=11, AT&T n=9; T-Mobile n=35), so it is reported as counts, never ranked. Walk-test first (Section 11).
The most important zone question for a capital decision is not where coverage is weak but whose problem the weakness is. A gate weak for every carrier at once is a building problem, and the venue pays. A gate weak for one or two carriers is a carrier-path problem, and the carrier's path pays. ONT's answer is unambiguous Derived:
Weak is a per-gate carrier mean 5G strength below −95 dBm; every per-gate carrier base is ≥ 30 tiles on the all-tile basis (5G-reporting bases disclosed where thinner — Section 10). Zero all-three-weak gates means the building itself propagates fine — Verizon proves it at every gate. The weakness travels with two specific carriers' paths, which routes the fix to carrier engagements rather than venue construction.
5.1The measurement resolves to 54 terminal-attributed tenants — concessions, retail, services, and transport/admin points from Google Places, each read against the per-carrier tiles within 80 m. The carrier story survives the trip down to the shop door intact: the weakest tenant readings in the venue are AT&T's, and they cluster in Terminal 4.
5.2The worst 15 tenant readings per carrier (mean 4G strength, ascending) are tabulated in Appendix B. The floor of the AT&T list sits at −107.6 dBm (Lot 4, Terminal 4) with the entire list at −104.8 dBm or weaker — every one of the 15 in Terminal 4. Verizon's worst list is dominated by outdoor parking lots and curb points, its weakest at −100.7 dBm; T-Mobile's worst tenant reads −96.7 dBm, barely below the healthy line. The same building, three different networks, visible from the storefront.
5.3Tenant categories are small cells — a handful of tenants each — so category-level comparisons would ride on fewer than 30 points and are not made. Tenant readings are reported individually, with their terminal, as observations rather than rankings.
Coverage is a static property of a building; load is what turns a marginal reading into a bad afternoon. The density layer counts unique Android devices per bin over the measurement window, deduplicated — it does not count simultaneous users and does not measure time spent. Read it as footfall reach, never as occupancy.
The two terminals run equal median density — 21 unique devices per bin in both footprints, roughly ten times the airport-wide median of 2 — so neither terminal "has more devices" than the other on the headline measure. The distribution tails differ: Terminal 2 peaks higher (p95 of 72 vs 56), consistent with its concentrated concessions core. International Arrivals is reported separately as a disclosed count: median 11 (n=36 tiles).
Crossing each zone's worst-carrier below-healthy share of 4G tiles with its median density ranks where weak signal meets device concentration Derived · labeled index, not a headcount: Terminal 4 first at ~1,862 (AT&T below healthy on 88.7% of T4 footprint 4G tiles × median 21), Terminal 2 second at ~1,638, then T2 apron/curb ~224 and T4 apron/curb ~184. T4 leads because AT&T's below-healthy share is higher there, not because it has more devices — the two terminals' medians are equal. International Arrivals is excluded from this ranking (its worst-carrier share would ride sub-floor per-carrier bases) and appears above as a count only. T4 remains where remediation buys the most exposure reduction per dB.
The usage layer records Wi-Fi and mobile megabytes at each tile, but it cannot tie Wi-Fi use to a specific carrier's subscribers. The locked rule: venue- and zone-level offload shares are defensible because the denominator is all networks; per-carrier offload is not derivable from this dataset and is not shown, implied, or estimated anywhere in this report.
Σ Wi-Fi ÷ Σ (Wi-Fi + mobile) observed megabytes over the window, volume basis, over the 16,267-tile usage layer. Venue and zone basis only.
Terminal 4 posts the venue's lowest offload at 34.1%, twenty points under Terminal 2's 54.3% and well under the airport-wide 45.9%. That is the Wi-Fi-side face of the same T4 story: the terminal with the weakest worst-carrier cellular is also the one where Wi-Fi relieves the least traffic. Whether the cause is coverage, onboarding friction, or capacity is not answerable from tiles — it needs controller telemetry, which is Recommendation 5.
Weak cellular with strong Wi-Fi offload is a managed problem; weak cellular with the venue's lowest offload is exposure stacking. Terminal 4 pairs terminal-grade device concentration (median 21, equal to T2's) with the weakest worst-carrier cellular and the least Wi-Fi relief — the same zone, three independent layers, one conclusion.
The airport graph records a Private 5G procurement at ONT at RFP stage Imagine knowledge graph. Public measurement does not characterize private networks, so this section describes the record and the decision it opens, not a deployment. The status should be verified against the live procurement record before external use.
The procurement matters to this report because it is the venue's natural moment to decide the public-carrier question at the same table. The measurement gives that decision its baseline: one carrier delivering through a registry-recorded DAS, one riding macro spillover at the edge of healthy, one weak on both sides of the skin. Whether the answer is a neutral-host or multi-carrier path bolted to the same program, or separate carrier-funded remediation, writing the acceptance criteria in Section 14 into the RFP costs nothing and anchors every later conversation.
A Private 5G network serves the venue's operational workloads; it does not fix a public carrier's passenger experience. Treating the RFP as the moment to also decide the public-carrier path — Recommendation 1 — keeps the two from being conflated and keeps the venue from buying one and assuming the other.
The bands a carrier's strongest-serving tiles attach to reveal which of its spectrum actually works in the venue. Verizon's 5G leans C-band (3.5–3.7 GHz) — capacity spectrum delivered indoors, which fits a fed DAS. T-Mobile leans its 2.5–2.6 GHz mid-band layer. AT&T's 5G barely registers: its band counts are so thin they are reported as counts only, never compared.
Tile counts, airport-wide, per strongest-serving band. AT&T's 5G bases (36 / 31 / 18 tiles) sit below the 30-tile comparison floor in two of three bands and are shown as counts only. Full tables in Appendix C.
Availability figures are labeled means over the tiles that report the field — an Ookla-defined availability score, not uptime — and absence of a reading is absence of measurement, never evidence of no service.
All three carriers show effectively universal 4G availability in the measured basis — a passenger always has a usable signal. ONT's problems are experience-grade, not outage-grade. The 5G strength and quality figures decide the experience above that floor. AT&T's own-network mean of 96.1% is the one number below 99, consistent with its thin footprint.
A separate T-Mobile 5G raster render (~4 m/pixel, presence only — the color legend is not held, so no strength values are read from it) shows samples across 93.1% of the Terminal 2 footprint, 90.3% of Terminal 4, and 74.1% of International Arrivals. It corroborates the tile layer's sampling basis: T-Mobile's footprint is genuinely wide, and International Arrivals is genuinely the thinnest-sampled building.
Some gates work well on one carrier and poorly on another. The spread between the best and worst carrier mean at a gate, in dB, quantifies how much a passenger's experience depends on which network they happen to carry Derived. The ranked atlas admits only gates where every carrier's 5G-reporting tile base clears the 30-tile floor — which excludes T4 gates 401 and 402 (AT&T 5G-reporting n=15 and n=25). On that basis the atlas still has one author: across the twenty widest ranked gates, the worst carrier is AT&T at all twenty, with spreads from 13.5 to 25.9 dB. The five widest are all Terminal 4 gates: 403 (25.9 dB), 413 (24.8), 414 (24.6), 407 (24.5), 412 (24.5).
Gates 401 and 402 show the largest apparent spreads in the venue — 31.7 and 27.5 dB — but AT&T's 5G-reporting bases there are n=15 and n=25, below the 30-tile floor, so they are excluded from the ranking and from the Section 14 criteria and reported here as indicative counts only. Establishing a floor-clearing measurement basis at these two gates is itself a Section 14 line item.
Spread = best minus worst per-gate carrier mean 5G strength, ranked over gates where every carrier's 5G-reporting base is ≥ 30 tiles (gates 401/402 excluded as sub-floor, disclosed above). Per-gate n in Appendix A. Worst carrier is AT&T at all 20 ranked gates; floor of the list 13.5 dB at T2 gate 205.
A venue that hosts every carrier well makes them perform alike at the same gate. Compressing the best-to-worst spread is the single clearest, most measurable success criterion for any remediation — easy to test before and after, hard to argue with. Section 14 sets the target: the top-5 gates from over 24 dB to ≤ 10 dB.
ONT's federal footprint in the public data is the International Arrivals building and its U.S. Customs and Border Protection operation. Two disciplined statements are all the data supports. First: the IA building is the weakest-measured building at the venue, and also the thinnest-measured — per-carrier 5G bases fall below the 30-tile floor for two of three carriers (Verizon n=11, AT&T n=9; T-Mobile n=35), so per-carrier results there are reported as counts and never ranked. Second: the CBP tenant point reads at-risk on AT&T (−99.1 dBm 4G) and healthy on Verizon (−85.2) and T-Mobile (−92.6) — a single Places point, reported for scope, not as a finding.
AT&T's own band table (Appendix C) shows 700 MHz public-safety spectrum — Band 14, the block licensed to FirstNet and carried by AT&T — as the strongest serving 4G band on 234 AT&T tiles at the venue. Read that against Section 3: the carrier that is weakest at every one of 27 gates is the one public-safety users ride. CBP, TSA, and first-responder devices on FirstNet inherit AT&T's venue profile. A public-safety-grade coverage review belongs in the scope of any in-building program here, and this finding is the authority's strongest lever for AT&T-funded remediation.
The federal hall is exactly where a crowdsourced layer is structurally thinnest: controlled space, devices stowed, dense construction. The honest read is that public measurement cannot characterize this building, and the first action there is a walk-test baseline — Recommendation 6 — not a remediation design drawn from nine tiles.
dBm does not move a budget; passengers do. Crossing ONT's 3,494,554 annual enplanements Imagine knowledge graph with each carrier's gates below the healthy 5G line translates the RF picture into the unit an airport funds against Derived. One assumption, stated and labeled: boardings are spread uniformly across the 27 gates Assumption · order-of-magnitude proxy, not a measurement.
Read plainly: under the stated proxy, every passenger who boards at ONT on AT&T does so at a gate below the healthy 5G line, and a T-Mobile passenger does so about four times in five. The same arithmetic re-run after remediation, on the same gates, is the return-on-investment measure. A gate-level enplanement feed replaces the uniform proxy when available (Section 15).
Seven actions, ordered by how much they change the venue's outcome. Each is grounded in the measurement record, with engineering inference labeled as such.
The RFP Imagine knowledge graph is the venue's natural decision point. Decide deliberately whether the path is neutral-host, a multi-carrier extension, or carrier-funded remediation — rather than letting the public-carrier gap ride unaddressed alongside a private-network award.
Owner · OIAA procurementThe system is doing its job in the terminals; the record of it is a cited KG edge, not a held document. Document the as-built — coverage zones, whether IA is on it, expansion and multi-carrier capacity — before any procurement decision assumes what it can or cannot host.
Owner · OIAA with VerizonSection 14 is drafted to be pasted: per-gate thresholds, the T4 remediation target, the spread-compression test. Criteria written before award cost nothing; criteria written after cost negotiations.
Owner · OIAAWeakest 4G at 27 of 27 gates, below-healthy 5G at 27 of 27, weak on both sides of the skin, worst in T4 — this is a carrier-path problem with a measurable baseline, and the density-weighted view says T4 remediation buys the most exposure reduction. The evidence package is this report.
Owner · OIAA with AT&T34.1% offload against Terminal 2's 54.3% is a twenty-point gap the tile layer can see but not explain. Access-point placement, onboarding friction, or capacity — controller data answers in a week what tiles cannot.
Owner · OIAA ITThe weakest-measured and thinnest-measured building. A scanner walk-test establishes the baseline that nine crowdsourced tiles cannot, and it is the prerequisite for any federal-stakeholder conversation about the hall.
Owner · OIAA with survey vendorSection 15's list — sample counts per bin, indoor mask, date-range comparison, cell-site locations — each closes a named caveat in this report and sharpens the re-measurement plan in Section 14.
Owner · Imagine WirelessProposed acceptance tests, written against this report's baseline and measurable with the same public layer that produced it. Thresholds are the platform ladder from Section 16; the criteria are recommendations, stated as such, for OIAA to adopt or tighten.
| Criterion | Zone | Carrier | Threshold | Baseline today |
|---|---|---|---|---|
| Every hosted carrier healthy at every gate | All 27 gates | All | Per-gate mean 5G strength ≥ −95 dBm for every hosted carrier | Verizon 27/27 · T-Mobile 6/27 · AT&T 0/27 |
| No gate near the critical band | All 27 gates | All | No per-gate carrier mean below −110 dBm | Held today · keep it held |
| AT&T Terminal 4 lifted | Terminal 4 | AT&T | Below-healthy share of footprint 4G tiles from 88.7% to ≤ 25% | 88.7% below healthy · n=141 |
| T-Mobile quality above the mark | Terminal footprints | T-Mobile | Median 5G quality from +5 dB to ≥ +10 dB | +5 dB · n=331 |
| International Arrivals baselined | Intl Arrivals | All | Walk-test baseline established before any remediation design | 5G bases below the floor for two of three carriers · VZ n=11, ATT n=9 (TMO n=35) |
| Asymmetry compressed | Top-5 ranked spread gates | All | Best-to-worst carrier spread ≤ 10 dB at each | 25.9 / 24.8 / 24.6 / 24.5 / 24.5 dB at T4 gates 403 · 413 · 414 · 407 · 412 |
| Gates 401 / 402 measurable first | Terminal 4 | AT&T | Establish a floor-clearing measurement basis (walk-test or re-pull) before setting a spread target — apparent spreads 31.7 / 27.5 dB ride on 5G-reporting n=15 / 25 | Sub-floor · indicative only |
Re-measurement plan. Re-pull the same tile layers on the same polygons at remediation plus 90 days and annually after, adding the layers this edition lacks (Section 15). The asymmetry table in Appendix A, re-run on the same gates, is the before-and-after exhibit for the board.
| Next layer to add | What it would resolve | Source |
|---|---|---|
| Sample counts per bin | An n column on every tile, closing the crowdsource credibility gap below the current zone-level n discipline | Ookla Imported Points Report |
| Native indoor / outdoor mask | Replaces the footprint-polygon segmentation behind the fingerprint diagnostic | Ookla portal re-pull |
| Date-range comparison | Before-and-after for any remediation, painted improvement / degradation natively | Ookla portal |
| Cell-site locations | Shows which macro sites feed the terminals, turning the thin-grid inference into a map | Ookla portal |
| Indoor walk-test | International Arrivals baseline; jet bridges and back-of-house, out of tile scope | Scanner survey |
| Wi-Fi controller telemetry | Explains Terminal 4's 34.1% offload; airtime, client counts, per-AP capacity | OIAA IT |
| Verizon DAS as-built | Documents coverage zones, IA scope, and multi-carrier capacity of the working system | Verizon, OIAA |
| Gate-level enplanements | Replaces the uniform-distribution proxy in the Section 12 translation | OIAA operations |
| Speedtest performance layer | Throughput and latency — an experience metric above the coverage floor, which this edition does not carry | Ookla |
The platform holds per-gate detail behind every figure here: all 27 gate rows, worst-15 tenants per carrier, complete band tables. We'll walk your team through any zone in the venue.
Source and geometry. Ookla Cell Analytics tile dataset for ONT over a twelve-month window (2025-06-01 to 2026-05-31), 78 days fresh at issue: 29,205 per-carrier tile-aggregates and 16,267 usage tiles after clipping to the aerodrome. Bins are coordinate-rounded at 0.0001°, a measured pitch of roughly 11 × 9 m, with no Z-axis. Tiles are summarized onto terminal footprints and 80 m gate envelopes from a pinned OpenStreetMap snapshot and tenant points from Google Places. Every headline figure is a median; means appear only where labeled (sub-1 GHz reliance, service availability).
Threshold ladder, one everywhere. Strength healthy ≥ −95 dBm, at-risk −95 to −110, critical below −110. Quality healthy ≥ +10 dB. The same ladder governs every class share, gate count, and acceptance criterion in this report; no second weak-signal line is used anywhere. One simplification, stated: the ladder applies a single RSRP line-set to both 4G RSRP and 5G SS-RSRP, which are measured against different reference signals and are not strictly equivalent at equal experience. That is a deliberate platform simplification, applied identically to all three carriers; a venue-specified ladder would override it and be named here.
Minimum sample rule. No chart or comparative claim rests on a cell below n = 30 field-reporting tiles or points, enforced at build time; the floor governs tiles per cell, and per-gate or per-tenant diagnostic tables are published with per-row n. Sub-floor cells — International Arrivals per-carrier 5G (Verizon n=11, AT&T n=9; T-Mobile n=35 clears it), tenant categories, AT&T's 5G band counts, and AT&T's 5G-reporting bases at gates 401/402 (n=15/25) — are reported as counts, marked indicative, and never ranked. Gate-level results are stated as counts over 27 gates (27/27, 21/27) with each per-gate carrier base ≥ 30 tiles on the all-tile basis.
Panel composition. The Ookla panel samples Android devices only; iOS devices are not sampled, which leaves roughly half of passenger handsets invisible to this dataset. Carrier comparisons therefore assume Android and iOS devices experience the same radio network. The enplanement figure (3,494,554) is the latest FAA calendar-year figure carried in the registry snapshot, captured 2026-07-02 Imagine knowledge graph.
Confidence and provenance. Facts from the airport knowledge graph carry the inline tag Imagine knowledge graph — cited, not held as primary documents, and never the basis of a measured claim. Derived figures are tagged Derived; stated assumptions Assumption. The carrier ordering throughout — Verizon, T-Mobile, AT&T — is computed strongest-indoor-first, never hand-ordered.
Locked metric definitions. User density is the count of unique Android devices observed per bin over the window, deduplicated — it does not count simultaneous users and does not measure time spent. Offload is Σ Wi-Fi ÷ Σ (Wi-Fi + mobile) observed megabytes, volume basis, computed at venue and zone level only; per-carrier offload is not derivable from this dataset and is never implied. Service availability is a labeled mean over reporting tiles — absence of a reading is absence of measurement, not evidence of no service.
Caveats, stated precisely. This export carries no throughput, latency, or jitter; no such figure appears. The T-Mobile raster layer is presence-only (~4 m/pixel; the class-to-dBm legend is not held, so no strength value is ever read from a color). The sub-1 GHz metric carries a disclosed export artifact: 51 of 28,775 tiles report a share above 100%, excluded from the cleaned Section 3.4 means and disclosed alongside the data-layer values that retain them. The Verizon DAS and the Private 5G RFP are knowledge-graph records, not held as primary documents; the measured carrier asymmetry stands without them. No value in this report was estimated or invented; where a layer is missing, the section says so and ships no number.
Per-gate per-carrier means over tiles within the gate envelope; n = that carrier's tile count at the gate on the all-tile basis (all ≥ 30, build-enforced). 5G-reporting bases can be smaller: ◆ marks the two indicative cells — AT&T 5G at gate 401 (5G-reporting n=15) and 402 (n=25), below the 30 floor and excluded from ranked views and Section 14 criteria. Spread = best minus worst carrier 5G mean. Worst carrier is AT&T at all 27 gates on 4G and at every ranked spread gate.
Mean 4G strength over per-carrier tiles within 80 m of the Google Places tenant point, ranked ascending; 54 terminal-attributed tenants in the basis. Individual tenant rows are observations at the ~11 × 9 m bin floor, not rankings between tenants.
The interactive view is Fig. 6 in Section 9. Tile counts by band, both generations; AT&T 5G rows are thin bases (< 30 tiles in two of three bands), disclosed as counts only: