A campus network is judged on its worst hour, not its average week.
That hour is a full hall at the start of a timed assessment, or the morning results go up. Campus Wi-Fi planning at a Nigerian university has to start from that hour and work backwards. A network that feels fast at a tenth of its occupancy is a different network from the one that has to carry every device in the room at once.
Coverage is not capacity
Most campus wireless is designed for coverage. Someone walks the building, confirms there is signal everywhere, and signs it off. Coverage answers one question: can a device see the network? During an exam the question is different. Can every device in this room transmit without the airtime running out?
Radio is a shared medium. Devices on the same channel take turns to transmit, so as the number of active clients on one radio rises, each waits longer for its turn. Throughput per device falls and retries climb. Students do not report "no signal". They report a page that will not load and a submission that timed out.
That is why a network that has been fine for years can fail on its most visible day. Nothing broke. It was never built for that load.
First, decide whether the exam belongs on Wi-Fi
Say this before buying a single access point: if you can put a timed exam on a wired lab, do. A cabled computer-based test centre with a UPS behind the switches takes airtime and the login rush out of the problem entirely. Most of the work in this article exists because a hall full of students' own laptops has to be served over the air.
So there are two honest designs. One is a wired exam room and Wi-Fi for everything else: the library, registration, results day, lectures. The other is a hall where students sit the exam on their own devices over Wi-Fi, which is the harder case and the one the rest of this piece plans for. Pick deliberately. Many institutions end up in the second by accident, because a hall built for lectures got used for an exam.
Count the devices you will actually see
Capacity starts with an honest number: how many devices will be connected and active in each space at peak. Count generously. Four hundred students who each carry a phone and a laptop is up to 800 devices. A phone left in a pocket still connects and uses airtime, even if nobody touches it.
Do this per room, for the rooms that carry the peak: exam halls, lecture theatres used for tests, the library in revision weeks, and wherever registration happens. For each, write down the device count and what those devices will be doing. A browser-based test platform behaves differently from a hall streaming a lecture, so base the throughput per device on the heaviest use you expect in that room.
The output is a target for each room: devices times throughput per device, which gives the number of radios you need. Buy after that, never before. Buying first is how a campus ends up with expensive equipment that is still short in the three rooms that count.
More radios, less power
High density is solved by putting more radios closer to the users, each serving a smaller area. The instinct to turn up the transmit power so one access point reaches the whole hall makes things worse. Every device hears every access point, and channels can no longer be reused across the room.
The pattern that works in a large hall is several access points at lower power, each serving a defined block of seats. That needs a design and an on-site survey, not a guess from the floor plan. Nigerian campus buildings make the survey non-negotiable. Reinforced concrete and blockwork absorb and reflect signal far more than the plasterboard partitions much design guidance assumes, so a predictive model has to be checked in the real room.
Channel planning is the other half. The 2.4 GHz band has only three non-overlapping channels, which interfere badly at density, so treat it as a fallback for old devices. Put the load on 5 GHz, which has many more. Use narrower channels in dense rooms: more independent cells usually matter more in a packed hall than the headline speed of a wide channel.
There is also new room to work with. Since February 2026 the Nigerian Communications Commission's framework allows licence-exempt Wi-Fi in the lower 6 GHz band, 5925 to 6425 MHz, indoors at low power and on NCC type-approved equipment. Only Wi-Fi 6E and Wi-Fi 7 devices can use it, so it relieves the 5 GHz band rather than replacing it. Read the framework's conditions before you plan a hall around it.
The wire behind the radios
Wireless capacity is wasted if the cable behind it is the bottleneck. Each access point needs an uplink sized for what it will carry, and the switches feeding an exam hall need the uplink headroom to carry every access point at full load at the same time. Plan Power over Ethernet as carefully as data. Many current access points need PoE+ (802.3at, up to 30 W per port) to run all their radios, and the switch has a total PoE budget that every port draws from.
The internet link depends on where the exam platform lives. A platform hosted on campus keeps most of the traffic inside. A cloud platform sends every student's session out through the campus uplink, and that link becomes the single point of failure for the whole exam. Size it for concurrent sessions at peak. Fibre cuts are a routine operating fact in Nigeria, so put the second path on a different carrier and, where you can, a different physical route.
None of it survives a grid outage without protection. Every switch between the core and the exam hall needs UPS cover with enough runtime to ride through the changeover to generator. Wireless capacity planned in detail, with the switch closet on raw mains, has only moved the point of failure. Our guide to power protection and UPS planning covers the sizing.
Plan for the login rush
The first thing several hundred students do when they sit down is connect. If every device tries to log in within the same few minutes, authentication becomes the bottleneck before a single exam page loads. A captive portal that asks each student to type credentials into a browser will queue and time out, and a wave of failed logins looks exactly like an outage.
Use WPA2-Enterprise or WPA3-Enterprise with at least two RADIUS servers behind it. Devices that have connected before reconnect without anyone typing anything, which flattens the spike. Get students onboarded in the weeks before the exam, not at the door. Put the authentication servers on the same protected power as the network core, because a RADIUS server that drops in an outage takes the whole hall offline however well the radios were planned.
Keep exam traffic on its own SSID and VLAN, apart from general student access and staff systems. The wider segmentation of a campus network, including CCTV and access control, is covered in our guide to IT infrastructure for schools and universities. Student records and exam results are personal data under the Nigeria Data Protection Act 2023, so the access controls and logging around those systems should be designed in from the start.
Rehearse before the real thing
A design is a guess until it carries real load. A coverage walk with a signal meter proves nothing about capacity. Run a mock exam in the actual hall with real devices and real students, weeks before the live one. It shows where the airtime and the uplink run out while there is still time to fix them.
On the day, someone should be watching the hall live: client counts per access point and login success. A problem caught in the room can be handled. One reported afterwards becomes a disputed result. Write down what saturated and what held, and use it for the next capacity decision.
Frequently asked questions
How many access points does a large exam hall need?
There is no single number. It follows from the device count and the throughput each device needs on your exam platform. A 400-seat hall where students bring a phone and a laptop can present up to 800 devices, which means several access points at controlled power serving defined blocks of seats, not one or two powerful ones. A site survey in the real room confirms the count.
Is a coverage survey enough to show the network is ready for exams?
No. Coverage shows that devices can see the network. It says nothing about whether hundreds of them can transmit at once. Readiness is shown by a mock exam with real devices in the actual hall, or by load testing that simulates many active clients.
What usually fails during a computer-based test on Wi-Fi?
Rarely the radios themselves. The usual cause is something behind them: an authentication service that cannot take a simultaneous login surge, or a single point of failure such as one RADIUS server or a switch on unprotected power. An internet link sized for an average day fails the same way. Each of them looks like a network outage to the students in the hall.
Can we use the new 6 GHz band in Nigeria?
Yes, the lower part of it. The NCC's framework opens 5925 to 6425 MHz for licence-exempt indoor Wi-Fi on type-approved equipment. Only Wi-Fi 6E and newer devices can use it, so plan for 5 GHz first and treat 6 GHz as extra capacity for the devices that support it.
Where to start
Pick the one room that carries your heaviest peak. Write down its seat count and where the exam platform is hosted. Those two facts are enough for a first capacity estimate. We survey and design campus networks as part of our infrastructure solutions work for the education sector. Send us those two facts and we will tell you whether that room should be wired or wireless, and whether the network it has today would hold.
