Checking a carrier's coverage claim with FCC broadband data

When the gap is the whole argument

Most applications for a new tower rest on one claim: there is a gap in service, and this site closes it. The applicant brings propagation maps that show red where service fails today and green where the new site would fix it. The board has to decide how much weight those maps deserve, and for years it had almost nothing independent to test them against. The FCC's Broadband Data Collection changed that. Facilities-based mobile providers now file modeled coverage with the federal government twice a year, and the filings are public.

So when an applicant says a neighborhood lacks reliable 4G LTE, the town can open the provider's own federal filing and see what the provider told the FCC about the same streets. That comparison does not decide the application. It does something more useful: it tells staff which questions to ask, and it puts a second, independent model into the record.

If the federal filing and the application tell the same story, the record gets stronger for everyone, the applicant included. If they disagree, the applicant should explain why, and there are often good reasons. Either way the board ends up with findings built on evidence it can describe in plain words, which is what a reviewing court will look for later.

What a mobile provider files with the FCC

Under the Broadband Data Collection, each mobile provider submits propagation-modeled coverage maps for every technology it offers, including 4G LTE and 5G-NR. Each map shows where the provider's model predicts a user can reach a stated minimum download and upload speed. The FCC sets the speed tiers and the key modeling parameters, including the probability of reaching the speed at the edge of the cell and the load assumed on the cell, so every provider's filing is built on the same assumptions.

Providers file separately for two environments. Outdoor stationary coverage predicts service for a person standing still outside. In-vehicle mobile coverage predicts service for a person moving in a car, which is a harder standard because the signal has to pass through the vehicle body and hold through handoffs between sites. A street can be covered in the first layer and absent from the second.

That distinction matters more than anything else in this article, because applicants, residents and board members usually mean different things when they say "coverage." Write down which environment each person is talking about before you compare anything.

Indoor service is not a layer in the federal filing. Many tower applications that argue a gap are really arguing about in-building service, where people make calls from kitchens and offices. The federal data cannot confirm or refute an in-building claim directly. It can still show whether the outdoor signal that in-building service depends on is present, and that narrows the argument considerably.

The FCC publishes the filings on the National Broadband Map and as downloadable files. The public map summarizes mobile coverage on a hexagon grid, which makes it easy to read at neighborhood scale and easy to join to other data. Each download comes with documentation that states the technology, speed tier, environment and as-of date for every layer. Read that documentation before you map anything.

Four kinds of evidence, and what each one proves

A coverage dispute usually involves some mix of the four sources below. They answer different questions, and most confusion in a hearing comes from treating one as if it answered another's question.

Evidence	Where it comes from	What it shows	Where it falls short
FCC filed coverage	The provider's twice-yearly federal filing, public on the National Broadband Map	Modeled outdoor stationary and in-vehicle coverage at stated speeds, built on standard federal parameters	Modeled rather than measured, no indoor layer, and it trails the network in the field by months
Applicant propagation maps	The applicant's RF engineer, prepared for this application	Predicted signal strength before and after the proposed site, often drawn at an in-building threshold	Assumptions are chosen by the party seeking approval, and thresholds vary from one filing to the next
Drive tests	Test equipment driven along a route by the applicant or the town's consultant	Measured signal at specific places and times	Covers only the roads driven, and results depend on equipment, antenna placement, time of day and network load
Crowd-sourced speed tests	Individual users running apps on their own phones	Real devices in real places, including indoors	Uneven sampling, unknown device placement, and heavy clustering where people happen to live and commute

None of the four is the truth on its own. The filed coverage and the applicant's maps are both models, often built with similar tools on different assumptions. The drive test and the speed tests are measurements, each with its own sampling problem. A finding is credible when two independent kinds of evidence point the same way, and it is fragile when it rests on one.

Getting the federal data in an afternoon

Start at the data download area of the National Broadband Map and select mobile broadband for your state. Choose the provider named in the application and download its layers for each technology, speed tier and environment. If the application concerns 4G LTE, pull the provider's 5G layers as well, since a claimed gap in one technology is sometimes covered by another.

Record the as-of date of every layer. A filing describes coverage as of a fixed date and is published some months later, so the newest public layer can lag the network by most of a year. If the applicant says a nearby site was decommissioned or added since then, that explains a difference, and the applicant should be able to document it with a site list and dates.

Pull the same layers for the other providers serving the area. Whether other networks' coverage matters to the legal question is for the town attorney, and the governing standard has shifted over the years. The data is still worth having so that nobody on the board is surprised by it at the hearing.

Then assemble the town's own layers: the municipal boundary, the road centerlines, parcels, address points, and the search ring the applicant describes. Put everything into one projected coordinate system suited to your area before you measure anything. Areas and distances computed in latitude and longitude degrees are meaningless, and it is the most common error in reviews done in a hurry.

The analysis, step by step

The review below fits inside a normal application schedule if it starts the day the application is complete. Start it any later and the shot clock will be running while you wait on files.

Step three is where reviews stall. Applicants often submit propagation maps as images inside a PDF. Ask for the underlying files, either vector polygons or a raster with the signal values, plus the threshold each color represents. If only images are available, georeference them to road intersections with a handful of control points, and say in the finding that the overlay is approximate.

Step four turns pictures into numbers a board can weigh. For the area the applicant calls a gap, report what share of it the federal filing shows as covered, separately for each environment. Then report road miles inside the in-vehicle layer and the count of address points inside the outdoor layer. Address counts are the number residents understand, because they map to homes and businesses.

Reading the applicant's propagation maps

A propagation map is a prediction from a model of terrain, ground clutter, antenna height, transmit power and frequency. It is not a measurement, and it is only as good as the assumptions fed into it. Ask the applicant for those assumptions in writing.

• The signal threshold for each color, and the service it represents (outdoor, in-vehicle or in-building)
• The frequency bands modeled in the before and after maps
• Antenna centerline height, azimuths and transmit power for the proposed site
• The terrain and clutter data sources and their dates
• The list of existing sites included in the before map, including sites in neighboring towns
• Whether the before map reflects coverage or capacity, and how capacity was estimated if it is the claim

The threshold is where most disagreements live. An in-building threshold is set well above an outdoor one, because the signal has to survive the walls. A map drawn at an in-building threshold shows a much larger gap than the same network drawn at an outdoor threshold, and both maps can be honest. The question for the board is whether the threshold matches the service the applicant says is missing, and whether the same threshold appears on the before and after maps.

Check the before map against the federal filing for missing sites. When the federal layer shows outdoor coverage in an area the before map paints red, one common explanation is that the before map omits an existing site, often one just across the town line. The site list resolves it in a day.

Measurements on the ground

A drive test is the best available check on any model when it is done well. Ask for the route, the dates and times, the equipment, where the antenna sat (roof-mounted or inside the cabin), the measured parameter, and the raw data file rather than only a colored map. For LTE, reference signal received power is the standard measure of signal strength, and it should be reported as the raw values the equipment logged.

Ask whether the test measured only the applicant's network or scanned every network on the route. Scanner-based equipment can log several providers in one pass, which gives the board context at no extra cost. Ask also whether the route covered the specific streets the applicant names as the gap. A drive test along the highway says little about a residential grid half a mile off it.

Crowd-sourced speed tests answer a narrower question: are real users in this area getting poor service? A cluster of failed or slow tests inside the claimed gap supports the applicant. A cluster of good results there is worth a written question. Be careful with commercial crowd-sourced datasets, since their licenses often restrict publishing them in a public record. Check the terms before a slide reaches the agenda packet.

A town can commission its own drive test when the stakes justify it. It is the most direct way to settle a factual dispute between the two models, and it puts measured data from a party with no stake in the outcome into the record.

What the federal data cannot tell you

The filed coverage is a strong reference point and a weak verdict. Know its limits before someone in the room points them out for you.

There is also a structural tension worth understanding without accusing anyone of anything. A provider's federal filing answers the question "where does our network meet the federal standard?" A zoning application answers "where does our service fall short of what customers need?" The two documents use different thresholds for different purposes, so a provider can file coverage with the FCC and claim a gap at the same address without either document being false.

If the town's analysis suggests the federal filing overstates coverage, the FCC's challenge process is the channel for that question, and it stays separate from the zoning decision. Keep the two tracks apart in staff reports so the application is judged on its own record.

Writing the finding so it holds up

Federal law requires that a decision denying a wireless facility be in writing and supported by substantial evidence contained in a written record. Courts read that record closely. The coverage analysis therefore has to be something a judge, an applicant's attorney and a resident can each read and reach the same understanding of what was measured.

The analyst's job is to describe what the evidence shows, not to decide the application. A finding that reads like an argument for one side will be treated as one, by the applicant if it leans against them and by residents if it leans the other way.

Build every finding from the same parts, in the same order.

• The question, stated in one sentence in the applicant's own terms
• Every data source with its as-of date, version and download date
• The technology, threshold and environment of each layer compared
• Area, road miles and address counts, with the method in a sentence each
• Where the sources agree, stated first
• Where they disagree, with the applicant's written explanation quoted
• The limits of each source, in plain words
• No recommendation on approval unless the board or staff asked for one

Presenting it to the hearing body

Lead with what the sources agree on. Boards trust an analysis more when it opens with common ground, and it shrinks the dispute to its real size. Then present each disagreement as a question with the applicant's answer beside it.

Use sentences a resident can repeat. An illustrative example: "The provider's federal filing shows outdoor coverage at the federal standard across most of the area described as a gap, but in-vehicle coverage on only part of it. The applicant's maps are drawn at an in-building threshold, which explains much of the difference." That paragraph does more work than any map.

Expect the question "so is there a gap or not?" The honest answer depends on which gap, and the analyst should say so by layer: outdoor, in-vehicle, in-building. Resist the pull to give a single yes or no when the evidence supports three different answers.

Hold the same line with everyone. If a resident cites the federal map as proof there is no gap, explain what the map does and does not cover, exactly as you would if the applicant cited it. Neutrality that only runs in one direction is not neutrality, and the record will show it.

Make the maps readable at the back of the room. Use a color-blind-safe palette, a legend in words rather than signal values, labeled streets, and the town boundary. Put the technical version with signal values in an appendix for the applicant's engineer and anyone else who wants it.

When the federal data is the wrong tool

Some applications turn on questions the filed coverage cannot reach, and forcing it into those reviews produces confident findings that do not survive the hearing.

• Capacity claims. When the applicant says existing sites are overloaded, the issue is congestion. The federal layers assume a standard load and cannot show it. Ask for traffic and utilization data from the provider instead.
• Specific buildings. A claim about service inside a hospital or a school needs indoor measurements at that building.
• Very small gaps. A valley or a single block can fall inside one hexagon. Use a drive test or walk test for anything smaller than the grid.
• Modifications to existing sites. A request that qualifies as an eligible facilities request under Section 6409(a) does not turn on coverage at all, so a coverage review adds nothing to it.

For everything else, the federal filing belongs in the file. It costs an afternoon, it comes from the provider itself, and it gives staff a fixed reference point before the first hearing rather than after the third.

Frequently asked questions