How this map is made

This map is a prediction, not a measurement. It calculates the signal a phone should receive, from the public list of antennas and a radio propagation model. Nobody went out to measure every street in Québec. This page explains how the calculation is done, what it can do, what it cannot do, and how we check it in the field.

In brief

What is calculated: the signal of a phone outdoors, 1.5 m above the ground, for each carrier, on a 100 m grid.
Where the antennas come from: the public database of Innovation, Science and Economic Development Canada (ISED), which lists the antennas declared by carriers (nearly 90,000 antenna sectors in Québec and neighbouring regions).
How the signal is calculated: an internationally standardized propagation model (ITU-R P.1812), applied to Québec’s actual terrain, including buildings and forest wherever laser surveys (LiDAR) exist.
What is not taken into account: the inside of buildings (only estimated with an average loss, selectable on the map), network congestion at peak hours, outages, and small antennas that carriers do not declare.
How much to trust it: compared with 956 points measured on the road, the map is off by about 8 dB on average, in either direction. That is about one class on our colour scale. A drive-test campaign is under way to check it elsewhere.

1. The source data

The antennas

Every carrier must declare to ISED the antennas it operates, with their position, height, orientation, tilt, frequency and transmit power. This data is public (Terrestrial Spectrum Licence Site Data Extract, Open Government Licence – Canada). We use the extract of August 31, 2026.

This raw data needs some preparation before it can be used:

Missing or inconsistent values. Some declarations omit the height, the tilt or the beam width, or give a power in an ambiguous unit. We fill them in with typical values for the carrier and the band, and every correction remains traceable.
Rooftop antennas. An antenna declared less than 3 m high is usually mounted on a roof. We then add the height of the building, read from the LiDAR survey.
Indoor antennas. Antennas installed indoors (shopping centres, metro stations) do not serve the street: they are left out of the outdoor calculation.
Shared antennas. Several carriers use the same antennas. Bell and Telus operate a shared network: a subscriber of one uses the antennas of the other. The map therefore combines them into a single “Bell / Telus” layer. Vidéotron uses some of Rogers’ antennas on certain bands. These antennas are declared under Rogers’ name only, and the map also attaches them to Vidéotron.
Beyond the borders. Antennas in Ontario, New Brunswick and Newfoundland and Labrador that cover Québec are included.

The terrain

Radio signals are blocked by hills, buildings and, to a lesser extent, forest. We use:

Québec’s LiDAR (airborne laser survey, metre-level accuracy), which captures the ground and what stands on it: buildings, trees;
where it does not exist, the national elevation model from Natural Resources Canada (MRDEM, 30 m).

The terrain is sampled every 20 m along each path between the antenna and the calculated point.

2. The calculation

The propagation model

We use recommendation ITU-R P.1812, an International Telecommunication Union model designed to predict the coverage of terrestrial services, through the crc-covlib library of the Communications Research Centre (Government of Canada). The model accounts for distance, frequency, the terrain crossed (diffraction over obstacles), the curvature of the Earth and the radio climate.

The calculated level is the one reached half of the time, at half of the locations in a given area: a typical value, neither the best nor the worst.

The grid

Québec is divided into 100 m cells. Each cell is calculated for every antenna in range.
The results are then grouped into hexagons of about 0.1 km² (about 350 m across), each summarizing 49 calculation points.
In total, about 1.7 million hexagons show at least one carrier, or about 180,000 km².

What a phone actually receives: the operational corrections

A first, purely theoretical calculation proved too optimistic compared with road measurements, by 7 dB on average. Four corrections bring it back to what a user actually gets:

Range capped at 14.5 km. Beyond that, even when the antenna’s signal still reaches the phone, it is the phone that can no longer be heard: it transmits much more weakly than the antenna. The network also limits the distance at which it accepts a device.
A −7 dB offset on all levels, measured from the road surveys.
Signal quality (SINR). A strong signal is useless if it is drowned out by neighbouring antennas of the same network. A point counts as covered only if the signal-to-(interference + noise) ratio reaches at least −6 dB, the minimum at which a 4G connection still works. The value shown is capped at 30 dB: beyond that, the quality of the transmitter and the phone limits the signal, no longer interference.
Roof height added to antennas declared too low (see above).

The best carrier

For each point and each carrier, the map keeps the antenna that offers the best usable signal. An antenna with a strong but noisy signal ranks behind a slightly weaker but clean one.

3. What the map shows

The quantity shown: RSRP

For 4G and 5G, the level shown is the RSRP (Reference Signal Received Power), in dBm. It is the value phones display in their technical menus, and the one test equipment measures. The closer the number is to zero, the better the signal.

Class	RSRP	What it usually means, outdoors
1 — excellent	−85 dBm and better	any use, including high-definition video
2 — good	−85 to −95 dBm	smooth browsing and calls
3 — fair	−95 to −105 dBm	works, at reduced speed
4 — weak	−105 to −115 dBm	calls and texts, slow or intermittent data
5 — marginal	−115 to −120 dBm	fragile connection, may drop
no coverage	below −120 dBm, or signal too noisy	no service expected

Reading a hexagon

Each hexagon summarizes 49 points. Two pieces of information are read together:

the level: that of the best point in the hexagon;
the covered share: the percentage of the 49 points where the carrier passes the thresholds.

A high level with a low covered share signals partial coverage: the edge of a coverage area, a valley, the shadow of a hill. The signal exists in the hexagon, but not everywhere.

The technologies

4G (LTE) and 5G: the main layer. It carries data today and, more and more, calls.
3G: a separate layer, for the two carriers that offer only 3G, Ice Wireless (Kuujjuaq, Quaqtaq) and i-Mobile (Gaspésie, Bas-Saint-Laurent, Anticosti). It is expressed in RSCP, the 3G equivalent of RSRP, but the two are not directly comparable: the same value does not mean the same thing in 3G and in 4G. The threshold there is −115 dBm. This layer does not account for interference, and the −7 dB offset, measured in 4G, is not verified for it. It is therefore less reliable than the main layer. The 3G of carriers that also have 4G is not shown: it is not calculated separately, and their 4G is what serves their subscribers.
5G at 3,500 and 3,800 MHz: not shown. These bands, with a very short range in cities, use antennas that steer their beams toward each user, which public data cannot describe. Where they exist, the map shows the coverage of the carrier’s other bands, which is what decides whether you have service.

On this site, 3G appears in the rankings with the label “3G only”. Since it carries only calls and slow data, and its prediction is less reliable, its score is multiplied by 0.6: at equal level, a 3G network ranks behind a 4G network. It is not included in the map’s “networks with good coverage” count, but it does count as service in areas “with no network”.

Carriers covered

Bell, Telus, Rogers and Vidéotron, as well as the regional networks: SSi Mobile (northern Cree communities), the Kativik Regional Government (Nunavik), Sogetel, Ecotel, Xplore, Orion Wireless and Cogeco. In 3G only: Ice Wireless and i-Mobile. Brands that use another company’s network (for example Fizz, Koodo, Virgin Plus, Fido, Lucky Mobile, Public Mobile) have the coverage of their host network: see the table of networks and providers.

4. What the map cannot do

The inside of buildings. The map predicts the signal on the street. Walls, and especially concrete, treated glass and metal, commonly remove 10 to 20 dB, sometimes more in a basement. A hexagon in class 2 can turn into class 4 deep inside a house, and nothing at all in an underground parking garage.

The inside of a vehicle. The car body usually removes 5 to 10 dB.

The situation selector. At the top right of the map, you can choose where you are: Outdoors (the prediction as is), Car (−7 dB), Indoors (−15 dB) or Deep indoors (−25 dB: basement, half-basement, core of a large building). The map, the network ranking and the scores then remove this average loss from the predicted signal. It is only an average: depending on the materials, the actual loss can be lower or higher. Xplore, whose antenna is mounted outside the house, is not attenuated. And the covered share of each hexagon remains the outdoor one: only its best point is known. For a trip, only the Outdoors and Car situations are offered.

Network load. A good signal does not guarantee good speed: at rush hour, in an arena or at a festival, the antenna shares its capacity among all its users. The map does not predict this.

Undeclared antennas. Small urban cells, repeaters, indoor antennas and temporary installations are not always in the public database. A recent antenna, or a poorly declared one, creates a gap we cannot see.

Point accuracy. The calculation is done on a 100 m grid, with terrain sampled every 20 m. A wall, an embankment or a grove smaller than that does not exist for the model. The map should be read at the scale of a neighbourhood, not of a precise address down to the metre.

Little-measured regions. Our verification measurements so far come from roads in southern Québec. How the map behaves in forests, mountains and the North has not yet been verified in the field.

5. How much to trust the map

What current measurements say

The map was compared with 956 hexagons measured in Lévis and along Autoroute 73 in southern Québec, with an analyzer whose antenna was placed outside the vehicle. Measurements too noisy to be reliable (RSRQ below −19.5 dB) were discarded.

	Bell / Telus	Rogers	Vidéotron	Best carrier
Median gap, measurement − map	−0.1 dB	+3.4 dB	+1.6 dB	+1.2 dB
Mean error, unsigned	8.9 dB	8.3 dB	8.2 dB	8.2 dB
Hexagons compared	865	736	712	1,042

A positive gap means the measurement is stronger than the map, so the map is conservative.

How to read these figures, honestly:

A median gap close to zero is not proof. The −7 dB offset was tuned on these same measurements. It is normal that they agree on average. The real check will come from new measurements, taken elsewhere: that is the purpose of the campaign below.
The figure that matters is the mean error of about 8 dB. At any given point, reality can differ from the map by one class, sometimes more. The map errs in both directions: slightly more than half of the measured points are better than predicted, the others worse.
The map sometimes misses real signal. In about 12% of the hexagons where Rogers or Vidéotron was measured, the map showed nothing for that carrier. About a third of these cases lie beyond the 14.5 km range. For Bell / Telus, such misses are rare. The map is therefore rather conservative about the extent of the Rogers and Vidéotron networks.
The sample is small and local: a single region, along roads, while driving.

For a decision that matters

Before relying on the map to stay reachable (hiking, isolated road, forestry work), keep one class of margin: treat a hexagon shown as “fair” as “weak”. Inside a building, keep two. And in an area where the map shows only a marginal signal, plan another means of communication.

6. The drive-test campaign

Why

A prediction is only as good as its verification. We are therefore launching a field measurement campaign, region by region, with three goals:

Measure the map’s error where it has never been checked: the North, forests, mountains, small towns, dense downtowns.
Find data errors: an antenna missing from the public database, an antenna in the wrong place or pointing the wrong way.
Correct the map, if the measurements justify it, in a public and traceable way.

What is measured

At each measurement point:

the GPS position and the time;
the carrier and the serving cell (cell identifier, band);
RSRP, RSRQ and SINR for 4G and 5G, RSCP for 3G;
the vehicle speed and the position of the measurement antenna (outside or in the cabin).

The campaign rules

Compare like with like. The map predicts the signal outdoors, at 1.5 m. A measurement taken with a phone in hand inside the car is corrected for the cabin loss, or analyzed separately.
Discard doubtful measurements. Measurements that are very noisy (RSRQ below −19.5 dB), taken without reliable GPS or at too high a speed are removed.
Group by hexagon. Measurements are pooled by map hexagon, and a hexagon is kept only with enough measurements to be representative.
Keep the measurements used for correction separate from those used for verification. A correction tuned on some measurements cannot be verified by those same measurements.

How the map will be corrected

No quiet touch-ups. Every correction is described on this page, with its date, the measurements that prompted it and its effect.
One correction per cause, not per point. If the measurements show, for example, that the map is too optimistic in dense forest, or for one carrier in one band, the rule is corrected everywhere it applies, not hexagon by hexagon. A map touched up by hand where measurements were taken would be right on the roads and wrong elsewhere.
A correction must hold elsewhere. Before it is adopted, a correction tuned on one region is checked on another. If it does not hold there, it is not applied.
Antenna errors are fixed at the source: antenna added, moved or reoriented in our data, with a note.
Every version of the map is dated and kept.

Taking part

How to take part will be published on this page: who can participate, with what equipment or app, how to send measurements and how they are protected and anonymized, and how to report a one-off discrepancy (“I have no signal here although the map shows some”).

Progress

Progress by region (hexagons checked, measured error by carrier, corrections applied) will be published here as the campaign proceeds.

Version	Date	Change
V1	October 16, 2026	first map, theoretical calculation
V2	October 31, 2026	operational corrections (range, −7 dB, signal quality, rooftops); 3G layer
V2 revised	November 2, 2026	signal quality (SINR) capped at 30 dB

Disclaimer

Coverage predicted by a model from ISED public data. It is not a measurement. It reflects neither network capacity nor load; inside buildings and vehicles, the signal is only estimated with an average loss. It is not an official carrier map. For a service question at a specific address, the carrier remains the reference.

Contains information licensed under the Open Government Licence – Canada. Source: Innovation, Science and Economic Development Canada, Terrestrial Spectrum Licence Site Data Extract. Propagation model: ITU-R P.1812, crc-covlib library of the Communications Research Centre. Terrain: Government of Québec LiDAR; MRDEM, Natural Resources Canada.