Building-wide Wi-Fi design starts with the cabling

Complaints about Wi-Fi in a multi-family building almost always arrive as a hardware question: which router, which mesh kit, which brand of access point. It is usually the wrong question. Building-wide Wi-Fi design is settled long before equipment is selected — by where cable was run, how far each drop sits from a switch, and what the building is made of.

Why stronger routers and mesh extenders fail at building scale

A more powerful access point does not extend usable coverage, because the limiting direction is the client's. An access point transmits harder than a phone can. Turn it up and the phone hears the access point from further away while the access point still cannot hear the reply. That is where full-bar dead zones come from.

Mesh extenders fail differently. A repeater that receives and retransmits on the same radio occupies the air twice for every packet, so throughput through the hop drops sharply, and a second hop drops it again. In a concrete building, the wireless backhaul is the weakest part of what you just bought.

Roaming is the third failure. Devices, not the network, decide when to leave an access point, so extenders with their own network name — or access points at mismatched power — produce clients that hold a poor connection while walking past a good one. The fix is more access points, each on a wired backhaul at moderate power, all advertising the same network.

Building-wide Wi-Fi design is an access point placement problem

Coverage is the set of places you were able to land a cable, because every access point needs one. PoE carries power and data over the same twisted-pair run, which is what makes ceiling mounting practical — but it consolidates the cable, it does not remove it.

That cable is bound by the 100 m channel limit for balanced twisted-pair horizontal cabling: conventionally 90 m of fixed horizontal cable plus up to 10 m of patch cords at both ends. The 90 m is cable path, not plan distance — a drop measuring 60 m across the floor plate can consume 85 m once it goes up a riser, along a tray and down to the device with service loops at both ends.

The consequence is structural. Telecom room positions decide which access point positions are reachable, and those positions decide coverage. On a long corridor building or a large floor plate, one central room does not reach the ends, and no equipment choice recovers from that. Getting the rooms right is part of the structured cabling design, not a networking afterthought.

Four specifics worth holding the design to:

• Cat6A rather than Cat6 for access point drops. It carries 10GBASE-T over the full channel, and its larger conductors run cooler in bundles under PoE load.
• PoE budget sized for every port loaded at once, not the per-port maximum. Tri-band access points commonly want PoE+ or better, and the same switches usually feed cameras and intercom.
• Mount below the ceiling, in the space being served. An access point in a ceiling void above metal ductwork, or in an electrical room because that is where the cable ended, is a wasted radio.
• Survey rather than guess. A predictive survey against real drawings with real wall types, validated on site after installation, shows where the holes are. A spacing rule of thumb gives you corridors full of interfering radios and suites with none.

What concrete, glass and the neighbours' networks do to coverage

The building

Poured concrete with rebar attenuates heavily. A slab behaves close to a hard boundary between floors, so any plan relying on signal bleeding between storeys will not hold. Metal studs, foil-faced insulation and metallized glazing each act as partial RF screens — a corner suite with glazed exterior walls and foil-backed sheathing is shielded in every direction. That is why rooftop radios rarely serve interiors, and why suite coverage means an access point inside the suite.

Elevators are their own problem: a cab is a metal box in a steel-lined hoistway. Coverage inside needs an antenna or access point in the shaft with its own cable, and coordination with the elevator contractor. Stairwells, garbage and storage rooms are the same problem in a different shape — concrete on all six sides.

The neighbours

In a dense multi-dwelling building, the main interference source is other people's networks, not microwave ovens. Every suite may have its own router, and Wi-Fi is a polite protocol — devices sharing a channel take turns, so co-channel interference costs airtime rather than signal strength. Partially overlapping channels are worse: they raise the noise floor and force retransmissions.

Two rules follow. Keep channel widths narrow: 80 MHz benchmarks well in an empty lab and leaves too few usable channels in a full building, so 20 or 40 MHz on 5 GHz gives better aggregate throughput. And fix the channel plan deliberately, reusing channels only where distance and concrete separate the access points; automatic RF management is a starting point, not a plan.

Amenity rooms, corridors and parkades are three different problems

One access point model at one spacing does not solve all three, because they fail for different reasons.

Amenity rooms are a capacity problem wearing a coverage costume. A gym, party room or co-work lounge packs a high device count and video and voice traffic into a small area, and it matters most on the one evening it is full — when the single access point covering it becomes the bottleneck. Design for airtime: more access points at lower power, narrow channels, 5 and 6 GHz preferred.

Corridors are where multi-family designs most often go wrong. Corridor access points are legitimate for staff, service devices and building systems, but never as suite coverage: a corridor acts as a waveguide, so the signal runs its length and stops at the demising wall. Suite Wi-Fi belongs on an in-suite access point fed by the suite's own drop.

Parkades need coverage pushed along drive aisles rather than spread from omnidirectional points, mounting heights that survive a truck, and enclosures suited to a wet, dusty space with cable in conduit. They also carry real load now: gate controllers, entry intercom, cameras and EV charger management. One thing worth telling residents plainly: Wi-Fi is not cellular.

Keep resident, guest and building systems on separate networks

Building systems must not share a guest network. Not a second network name on the same flat network — a genuinely separate one, with routing between segments denied by default.

Building systems means access control and door hardware, entry intercom, cameras and the recorder, elevator interfaces, mechanical controls, gates and EV chargers. These have to work while the guest network is saturated, being rebuilt, or has just had its password rotated. Three reasons to separate them: fault isolation, so a guest-side problem stops at the boundary; security, so device management interfaces are not reachable from a visitor's laptop; and predictability, since guest traffic should be shaped and door hardware should not be.

Wherever a device can take a cable, give it a cable. Wireless is for things that move, and almost nothing on that list moves. It is the same principle that drives access control rough-in.

Where cameras and access logs are involved, recording in a residential or workplace setting carries privacy obligations in BC. Who may view footage and door logs, and how long they are retained, should be settled with the strata council or owner and reviewed with legal counsel — not left to whoever holds the administrator password.

Resident traffic is a separate question. In most buildings, in-suite internet is the resident's own account on their own equipment, and the building's obligation is a clean pathway and a terminated drop. Managed resident Wi-Fi as an amenity is a different commercial commitment.

What to specify at pre-wire

The cheapest access point you will ever add is the one whose cable is already there. Adding a drop at rough-in costs cable and an hour. The same drop after occupancy costs ceiling access, drywall patching, painting, and a notice to residents.

• Cat6A home run to every planned access point location, terminated on a patch panel in the telecom room and labelled at both ends to a scheme that is written down.
• Spare drops at candidate locations you did not plan to use — corridor ends, amenity rooms, roof deck, storage rooms, parkade drive aisles.
• Telecom rooms positioned so no horizontal run exceeds 90 m of actual cable path. Measure the path, not the plan.
• Conduit sleeves with spare capacity at slab and rated-assembly penetrations, firestopped, so later additions do not mean coring.
• Power, cooling, grounding and UPS in every telecom room, plus a pathway from the demarcation point.
• Certified test results and as-built drawings at handover, showing final access point locations, the channel plan and where each drop lands. A building without as-builts is a building where the next contractor guesses.

Most of this is the same discipline that governs suite and riser cabling generally, covered in more depth in the multi-family pre-wire guide.

Where to start

If complaints are happening now, survey before buying. A validation survey of the occupied building separates a coverage problem from a capacity problem, an interference problem, or an undersized internet circuit — four different fixes, and only one is more access points.

If the building is still in design, put access point locations, telecom room positions and the pathways between them on the drawings before ceilings close. Orbit Automation is a licensed and insured low-voltage integrator in Surrey serving Metro Vancouver and the Fraser Valley; a site assessment is the usual starting point.