Wireless Site Surveys Toronto & GTA
Most bad Wi-Fi is not a hardware problem. It is a design problem that hardware was bought to solve. We measure what the radio is actually doing in your building, tell you the cause, and give you a plan you could hand to any contractor.
Tell Us What It's Doing. We'll Tell You Why.
Wireless problems get described in symptoms and diagnosed in guesses — usually followed by buying more access points. Pick the complaint that sounds like your building and jump straight to what is actually causing it.
Velocity Cabling surveys and designs commercial wireless networks across Toronto and the GTA — offices, warehouses and distribution centres, clinics and multi-floor buildings. This page is written as a diagnosis rather than a brochure, because in fifteen years of walking into buildings with bad Wi-Fi, the cause is nearly always one of the eight things above, and almost never the one the client expected.
Which Survey You Actually Need.
There are four, they answer different questions, and buying the wrong one is how organisations end up with a beautiful report that does not address the complaint. What decides it is not your building type. It is where you are in the project.
| Where you are | Survey | What it produces |
|---|---|---|
| Nothing installed yet, or a building still being built | Predictive | Modelled coverage from building materials and layout. AP count, placement, mount heights, antenna choice, channel and power plan, and the cabling schedule the electrical contractor can price from. |
| A network exists and people are complaining | Passive | Measurement of what the radio is really doing — signal, noise floor, signal-to-noise ratio, channel utilisation, co-channel overlap and every neighbouring network. Listens only, connects to nothing, changes nothing. |
| A design is installed and you need it proven | Active validation | Connects to your own SSID and measures throughput, retries, data rates and roaming behaviour as a person actually walks the space. This is the difference between a design and a working network. |
| It fails intermittently and nobody can reproduce it | Spectrum analysis | Looks below the Wi-Fi layer at the raw RF — the sources that are not Wi-Fi at all and are therefore invisible to every tool your IT team already owns. |
That last row is worth dwelling on. Wi-Fi shares unlicensed spectrum with things that are not Wi-Fi and do not take turns — certain cordless headsets and wireless cameras, some motion sensors, older microwave ovens, and the occasional piece of equipment radiating out of a failing seal. Wi-Fi analysers cannot see them, because those tools only report what identifies itself as a network. If your problem arrives on a schedule nobody can explain, this is usually why, and it is why the fault-finding survey exists as a separate engagement.
How an engagement runs
The Numbers We Design To.
Most wireless proposals describe an outcome — “strong, reliable coverage throughout.” That is unfalsifiable. These are the values we design and validate against, so the finished network either meets them or it does not, and you can hold us to it.
| What we measure | Target | Why it matters |
|---|---|---|
| Signal strength at cell edge (RSSI) | −67 dBm voice and scanning −70 dBm general data |
Below the target, devices fall to slower data rates, spend longer on air for the same work, and retry more. One weak device slows everyone on that radio. |
| Signal-to-noise ratio (SNR) | 25 dB or better for voice 20 dB for data |
Signal strength alone means nothing. What matters is how far the signal sits above the noise. A strong signal in a noisy room still performs badly. |
| Co-channel interference | No more than 3 access points on the same channel audible above −85 dBm | Every additional AP heard on that channel is another device your clients must wait behind before transmitting. |
| Channel utilisation at the busy hour | Under 40% | The single best predictor of user experience. Past roughly 50% the channel is congested and latency and jitter climb sharply — which is why the network feels fine at 8am and broken at 2pm. |
| Cell overlap | Roughly 15–20% between adjacent cells | Too little and devices hold a dying connection instead of moving. Too much and you have manufactured your own interference. |
| Minimum data rate | Legacy low rates disabled, floor set to suit the slowest device that must be supported | A device allowed to connect at the lowest legacy rate holds the channel far longer than a modern one doing the same task. |
| Retry rate | Into single figures under normal load | Retries are wasted airtime. High retries with strong signal almost always points at interference or the power asymmetry below. |
These are design targets rather than published standards — there is no regulation that says your Wi-Fi must reach −67 dBm. They are the values the industry has converged on for real-time traffic, and the reason we publish them is so you can compare quotes on something measurable. If another proposal will not commit to numbers, ask why.
Why Adding Access Points Made It Worse.
This is the call we get most often, and the answer surprises almost everyone: Wi-Fi is half duplex and shared. Within earshot of one another on the same channel, every device — including the access points themselves — must wait its turn before transmitting. The currency is airtime, not signal strength.
So two access points on the same channel, mounted where each can hear the other, do not give you twice the capacity. They divide the capacity that was already there, and add management overhead on top. Signal goes up, throughput goes down, and the obvious response — add a third — makes it worse again.
The 2.4 GHz band makes this unavoidable. In North America it offers only three non-overlapping 20 MHz channels: 1, 6 and 11. Put a fourth access point in a space where all of them are audible and something is reusing a channel, by arithmetic, no matter how good the hardware is. Every band beyond 2.4 GHz exists largely to escape this constraint — which is also why 2.4 GHz is best treated as a compatibility band for older equipment rather than the backbone of a modern design.
The fix is rarely more hardware. It is a channel plan, appropriate channel widths, and power set so that cells stop shouting over each other. We have improved throughput in more than one building by turning access points down and switching two of them off.
Full Bars, Nothing Loading.
A conversation needs both parties to be audible. An access point mounted on the ceiling, mains powered with proper antennas, can transmit far harder than the device in someone's hand — and a barcode scanner or an IoT sensor is weaker still, because its whole design brief is battery life.
Turn the access point up to maximum and you extend how far away a device can hear it. You do not change how far away the access point can hear the device. The result is a client showing a strong signal, associating happily, and then failing to move data, because its replies are not making it back. Every user reports the same thing: full bars, nothing loading.
Turning the transmit power up improves coverage.
It extends the downlink only, and simultaneously enlarges the cell so it overlaps neighbouring cells on the same channel — adding interference while creating one-way connections at the edge. Access point power should be planned against the weakest device class that has to work, then coverage filled with more access points at lower power.
2.4 GHz travels further, so it is the safer band for a big space.
It does travel further, and in a large facility that is the problem rather than the benefit. Longer reach means every access point hears every other one across three available channels, so the interference is self-inflicted. In most commercial designs 2.4 GHz is enabled on a minority of access points, purely to serve legacy devices.
A phone app showing signal strength is close enough to a survey.
A phone reports its own reception at one point in time, using an antenna and a radio nothing like your scanners. It cannot show noise floor, channel utilisation, retries, co-channel overlap or non-Wi-Fi interference — which is to say it cannot show any of the things that are usually causing the problem. It is useful for spotting a dead AP. It is not a measurement.
Newer access points will fix a bad wireless network.
New hardware in the wrong positions on the wrong channels produces the same problems with a larger invoice. Placement, channel plan and power are what determine whether a network works. Newer access points are worth buying — after the design, not instead of it.
Roaming Is The Client's Decision.
This catches out even experienced IT teams. Your access points do not hand devices to one another. The device decides, on its own, when to leave one access point for another — and the logic behind that decision belongs to whoever wrote the driver in the phone, the laptop or the scanner. Vendors implement it differently, and they change it in firmware without telling anybody.
The classic failure is the sticky client: a device that clings to the access point it first joined long after walking away from it, dragging the connection down to unusable rather than moving to the strong access point directly overhead. Wi-Fi calls break up. A scanner times out mid-pick. Nothing appears wrong on the controller, because from the network's point of view the device is connected.
You cannot fix this by ordering devices to behave. You fix it by designing an environment where the decision is easy — cell edges defined at a proper threshold, sensible overlap, power balanced across neighbours, and legacy data rates removed so a distant access point stops looking like a viable option. The 802.11k, 802.11v and 802.11r extensions help by giving the client a neighbour list, a nudge and a faster reassociation, but they assist a good design rather than substitute for one.
This is why the active survey matters. Roaming behaviour cannot be predicted from a coverage map. It has to be walked, with a real device, on your real network, along the routes people actually take — the corridor between the office and the warehouse, the path from reception to the loading dock, the stairwell people use instead of the lift.
Capacity, Not Coverage.
Almost every wireless network that fails in the afternoon was designed for coverage. Coverage asks “can a device hear an access point here?” Capacity asks “how many devices doing how much work can this access point actually serve?” They give completely different answers, and only one of them predicts how the network feels at 2pm.
A capacity design starts from the devices. How many are associated at the busy hour, not how many staff there are — count the phone, the laptop and the tablet separately, because the radio does. What each one needs, honestly, distinguishing a video call from an email client. Then a realistic per-access-point figure, well below the number on the datasheet, since that figure assumes a single modern client at close range with the channel to itself. Coverage is checked afterwards, and it is nearly always satisfied by then.
This is what separates spaces that fail from spaces that do not. An open floor with 200 people is a capacity problem. A training room that fills for two hours a week is a capacity problem with a schedule. A warehouse with forty scanners spread over 100,000 square feet is the opposite — a coverage problem with almost no capacity demand at all, where the cost of getting it wrong is a picker walking into a dead zone.
The Building Changes When You Fill It.
Warehouse Wi-Fi is where we see the most expensive mistakes in the GTA, and the reason is simple: the survey was done when the building was empty. That survey was accurate. It just described a different building.
Racking full of product absorbs and reflects signal in a way that empty racking does not. Liquids are close to the worst case, since water absorbs strongly at these frequencies — a beverage or chemical distribution facility can look completely different at full stock. Dense paper and card behave similarly. Metal reflects rather than absorbs, so aisles start acting like waveguides, carrying signal a long way down the aisle and almost none across it. A network that tested perfectly in an empty shell starts dropping scanners the week the racking fills.
Ceiling height compounds it. Mounting omnidirectional access points at twelve metres sprays most of the energy across the roof space and the top of the racking, when the scanners are at floor level in the aisles. In high-bay space we generally specify directional or patch antennas aimed down the aisles, so the coverage arrives where the work happens instead of where the mounting was convenient.
What we do about it. Model at realistic stock levels rather than as-built, survey at representative loading where the operation allows, and validate after go-live. Where a facility is seasonal, we say plainly which condition the design is guaranteed against. A warehouse survey that does not mention stock is a survey of an empty building.
Every Building Type Breaks Differently.
Signal passes through floor slabs more than people expect. An access point on channel 36 sitting directly above another on channel 36 interferes with it just as surely as one down the corridor. Channel plans have to be three-dimensional, and open stairwells, light wells and glass atria create vertical leakage paths that a floor-by-floor plan never accounts for.
Fewer walls means less natural separation between cells, so co-channel interference rises even though coverage looks flawless. These spaces usually want more access points at lower power on narrower channels — the opposite of the instinct. Glass partitions reflect more than they absorb.
Continuity matters more than peak speed. Shielded imaging rooms create hard RF shadows next to areas needing full coverage, and clinical devices roam constantly with staff. Corridors, stairwells and lifts are part of the coverage requirement, not an afterthought.
Large machinery reflects, moving equipment changes the environment through the shift, and drives and motors raise the noise floor. Spectrum analysis is usually part of the engagement rather than an option, because the interference is often not Wi-Fi at all.
Two networks in one building — point of sale that must never drop, and guest access that must never be allowed to starve it. Stock levels and seasonal fit-outs change the RF environment several times a year, so the design needs headroom rather than precision.
Common across older Toronto stock. Thick masonry, lath and plaster, and structural steel that was never documented. Predictive modelling is least reliable here, so these buildings are surveyed rather than modelled, and cabling routes usually constrain access point placement more than RF does.
6 GHz Is Different In Canada.
Most wireless advice online is written for the United States or Europe. On 6 GHz that matters, because Canada took a notably more open position than most of the world, and it changes what is worth designing here.
Canada released the full 6 GHz band for licence-exempt wireless local area networking, well beyond what European regulators have opened. Innovation, Science and Economic Development Canada sets the rules under RSS-248, covering low-power indoor operation, very-low-power devices — adopted in 2024 and permitted indoors and outdoors — and standard-power operation under the control of an automated frequency coordination system. Canada was also the first country in the world to approve an AFC operator for commercial service.
The practical effect for a Toronto building is room. 6 GHz offers enough spectrum to run wide channels without immediately recreating the channel-reuse squeeze that constrains 5 GHz, which is exactly what high-density spaces need.
Two things temper that. Higher frequencies attenuate faster, so a 6 GHz cell is smaller than a 5 GHz cell from the same access point in the same position — a building designed to lean on 6 GHz generally needs more access points, not fewer, and that has to be in the budget from the start. And spectrum you cannot connect to is not capacity: if the devices that matter to your operation are older scanners and legacy equipment, 6 GHz does nothing for them. We audit the device mix before recommending a band strategy.
Wi-Fi 6, 6E and Wi-Fi 7
Wi-Fi 7 access points are readily available for enterprise use, and in practice most deployments run multi-link operation across 5 GHz and 6 GHz rather than all three bands, given where client support and firmware maturity currently sit. The gains are real for dense and latency-sensitive environments. But the honest position is that generation matters far less than placement: a well-designed Wi-Fi 6 network beats a badly placed Wi-Fi 7 one every time, and we would rather spend your budget on the design and the cabling than on a generation your devices cannot use yet.
Wireless Networks Are Mostly Wire.
An access point is the last few metres of a wired network, and it is routinely where a good design comes undone. Two things are worth checking before anybody orders hardware.
Power. Full tri-radio access points frequently need 802.3bt rather than 802.3at. Where the switch cannot deliver it, the access point does not fail — it quietly runs in a reduced mode, disabling a radio or dropping a port to a lower speed. The network then underperforms for a reason that never appears in any wireless report, and people spend months blaming the RF design.
The cable. A modern access point can move more than a gigabit in aggregate, which puts the uplink rather than the radio at the bottleneck. That points at Cat6A to every access point location — and Cat6A is also what supports the higher PoE classes over distance without excessive heat rise in bundled cable. Running Cat6 to save a little now is the most common false economy we encounter on wireless projects.
We do both halves. Survey and design, then the structured cabling to each access point location, mounting and configuration — so there is no gap between the company that designed the wireless and the company that installed the cable feeding it.
Get the RF measured before you buy anything.
A survey costs a fraction of a wireless refresh, and it is the only way to know whether you need new access points or a new plan for the ones you have.
What You Receive.
A survey is worth what its documentation is worth. Everything below is included as standard, and it is written to be usable by whoever holds the network next — including a contractor who is not us.
| Deliverable | What it contains |
|---|---|
| Heatmaps | Signal strength, signal-to-noise ratio, channel utilisation, data rate and interference, per band, overlaid on your floor plan. |
| Access point schedule | Every location with mounting type, height, antenna selection and orientation, so the install is repeatable rather than improvised on the day. |
| Channel and power plan | Per access point, per band, planned in three dimensions on multi-floor sites. |
| Cabling schedule | Cable runs, lengths, pathway notes and PoE class per location — priceable by any contractor, not just us. |
| Findings and root cause | Written in plain language, with the measurements behind each conclusion, so it can be forwarded to a manager who is not an engineer. |
| Interference register | Non-Wi-Fi sources identified, located where possible, with what to do about each. |
| Native survey files | The raw project files, not only a PDF. Your data stays yours and a future survey can build on this one instead of starting over. |
| Validation survey | Repeated after installation, so the difference between the design and the delivered network is measured rather than claimed. |
That last-but-one row is deliberate. Handing over only a PDF is common in this industry, and it means the next survey starts from zero and has to be bought from the same supplier. We do not think that is a reasonable way to treat a client.
Wireless Survey Questions.
A survey measures how radio behaves inside your specific building, so access point placement, channel plan and power settings are based on evidence rather than a guess at spacing. You need one when coverage or performance already matters to the business — warehouse scanning, Wi-Fi calling, clinical devices, anything on a deadline. For a small open office with light use, a careful install without a full survey is often defensible, and we will say so rather than sell you a survey you do not need.
They answer different questions. A predictive survey models the building in software before anything is installed, and produces the AP count, placement and cabling schedule. A passive survey measures what the radios in the building are actually doing right now — signal, noise, channel utilisation and interference — without connecting to anything. An active survey connects to your own network and measures real throughput, retries and roaming behaviour. Predictive is for planning, passive is for diagnosis, active is for proof.
Because Wi-Fi is half duplex and shared. Devices on the same channel within earshot of each other take turns transmitting, so two access points on the same channel do not double capacity — they split the airtime that was already there. Adding hardware without a channel and power plan increases co-channel interference and reduces usable throughput. This is the single most common thing we are called in to unwind.
Usually transmit power asymmetry. An access point can transmit far harder than a phone, a laptop or especially a barcode scanner. Turn the AP up and the client sees a strong signal from a long way off — but the AP cannot hear the client's much weaker reply. The connection looks perfect on the device and barely functions. The fix is lower AP power and more of them, not the reverse.
It is driven by floor area, number of floors, environment and whether you need a design or a diagnosis. A single-floor commercial office survey typically starts around $800 to $1,500. Warehouses, healthcare and multi-floor buildings are quoted on square footage and complexity, because the survey time scales with the space rather than the headcount. We quote a fixed number before starting, not an hourly estimate.
Either. We run survey-only engagements for organisations with their own network team, and you receive the full deliverable set including the native survey files. We also do the whole job — survey, Cat6A to each AP location, mounting, channel and power configuration, and the post-install validation survey that proves the design landed.
Canada has more 6 GHz spectrum available than most of the world, which makes 6 GHz genuinely worth designing for here. The gate is your client devices — spectrum you cannot connect to is not capacity. The practical trap is power: full tri-radio Wi-Fi 6E and Wi-Fi 7 access points commonly need 802.3bt rather than 802.3at, and switches that cannot deliver it quietly run the AP in a reduced mode. We check the switch and the cable before recommending the AP.
Probably not. Racking full of product is a different radio environment from empty racking — liquids, metal and dense paper all absorb signal, and aisles become waveguides. A survey done at low stock will read optimistic and the network will fail at exactly the time it matters. We model or measure at representative stock levels and say plainly when a previous survey no longer describes the building.
Toronto, Mississauga, Brampton, Vaughan, Markham, Scarborough, Etobicoke, Oakville, Hamilton, Kitchener-Waterloo, Whitby and Pickering. Surveys in occupied buildings can be scheduled outside business hours where the environment allows it — though for capacity work we usually want to measure during the busy hour, not after everyone has gone home.
Toronto, The GTA & Southern Ontario.
Crews dispatch from our Toronto base and reach most GTA sites within an hour.
Find Out What It Is Before You Replace It.
Wireless surveys across Toronto and the GTA, booked within 48 hours. You get the measurements, the cause and a plan you could hand to anyone.
Mon–Fri 8am–10pm · Evening & weekend installs available · Toronto, GTA & Southern Ontario