About
Subscribe
  • Home
  • /
  • TechForum
  • /
  • How to design a WiFi network that works first time

How to design a WiFi network that works first time

By Waseem Bhamjee, Support Technician at Duxbury Networking
Johannesburg, 09 Oct 2025
Waseem Bhamjee, Support Technician at Duxbury Networking.
Waseem Bhamjee, Support Technician at Duxbury Networking.

Good wireless isn’t guesswork but requires careful planning. Before a single access point goes on a wall or a mast, we map the space, listen to the radio environment and test how real users will experience the network. That’s the difference between WiFi that “mostly works” and WiFi you can trust on a busy Monday morning.

Duxbury Networking's technical team employs a combination of methods because no single survey tells the whole story. We typically blend predictive design, passive surveys, active surveys and (in tricky sites) AP-on-a-stick validation. Used together, they provide a complete picture and a design that holds up once people and devices are in place.

Predictive design

This is where we start when time or site access is limited. Using floor plans and known building materials, we model how radio signals will move through the space. We account for wall types, heights, ceiling voids, racking, office density and the applications that will run on top (voice, video, scanners, point-of-sale, etc). A predictive design won’t replace going on site, but it gets you 70% to 90% of the way there and sets sensible expectations on AP counts, placement and cabling. It’s also a fast way to budget properly before anyone drills a hole.

Passive survey

A passive walk test is precisely that: we walk the site with survey tools that listen to everything in the air. We don’t join the network but measure it. This tells us where coverage is strong or thin, which channels are crowded, how much noise is present and where non-WiFi interference lives (think DECT phones, poorly shielded electronics or a neighbour’s AP turned up to 11). For live environments such as schools, warehouses and hospitals, this is the quickest way to see the RF reality on the ground.

Active survey

Of course, signal bars don’t tell you if a call will drop. An active survey connects to the WLAN and behaves like a real client. We measure throughput, latency, packet loss, roaming behaviour, DHCP/DNS responsiveness and how applications feel while you move. This is where we pick up issues that maps alone won’t show. For example, sticky clients, coverage that looks fine but can’t carry voice, or a mistuned controller turning a good design into a frustrating user experience.

AP-on-a-stick: Try before you mount

Some spaces won’t play by the rules on paper. High ceilings, long aisles, lots of glass, reinforced concrete or stacked pallets can all change how a signal behaves. With AP-on-a-stick (AOS), we mount a real access point on a tripod at the planned height and location, power it up and walk the area. It’s a quick way to prove (or disprove) assumptions before you run cables and commit to mounts. We lean on AOS for new builds, hotels, distribution centres and any site where a small test can save a big rework later.

How we put it together

A typical engagement looks like this:

  • Plan: Run a predictive design from accurate drawings, agree on coverage and capacity targets and identify risk areas.
  • Validate: Visit site; use passive and (if needed) AP-on-a-stick to confirm the model or adjust it.
  • Deploy: Place and tune APs, set channels and power, align SSIDs and security.
  • Prove: Finish with an active survey so we know roaming, voice and critical apps behave as expected.
  • Support: Schedule health checks or troubleshooting surveys when the environment changes, for instance, with new tenants, racks or neighbours. Even a spike in device count can have an impact.

That last point matters. A network that was perfect in March can feel different by October after a tenant next door installs their own WLAN or your operations team adds scanners and cameras. Periodic checks keep minor issues from turning into big ones.

Why this approach works

Each survey answers a different question. Predictive tells us what should happen. Passive tells us what is happening. Active tells us what users feel. AP-on-a-stick lets us test the odd corners where theory and reality don’t quite line up. When we layer those views, the design becomes simple to defend and easier to support. You get the correct number of APs in the right places, cabling where it’s useful and settings that match the way the site actually runs.

If you’re planning a roll-out or living with WiFi that drops at the worst possible time, Duxbury Networking's team can help you pick the right survey mix and turn it into a plan you can build against. No surprises or guesswork required. You get a network that does what it’s meant to do: connect people and devices without fuss.

Share

Editorial contacts

Karien Wood
Duxbury Networking
(+27) 011 351 9800
kwood@duxnet.co.za