How Many Access Points Does a School Need? Corridor or Per-Room

Why does a corridor access point pass the survey and then fail at 09:00?

A school survey is usually taken in an empty building during the holidays. It measures signal strength in a room nobody is using: the corridor unit reads comfortably at the back row, the report says the floor is covered, and term starts three weeks later.

That corridor unit is now serving two or three classrooms at once. Thirty students at two devices each means sixty to ninety devices on a single cell, all opening the same cloud-synced lesson in the same minute. Nothing has failed; the airtime has simply run out. Coverage asks whether a device can hear the access point. Capacity asks how much airtime is left once everybody is talking, and in a school the second question decides the layout.

Four things make a school harder than an office, and none of them appear on the architectural drawing.

• Lessons are synchronised by the bell, so traffic arrives in bursts rather than spreading across the day
• Seating is fixed, and the far corner behind the door is the worst seat in every lesson, every day
• A full room is not transparent: thirty seated students absorb the signal the empty survey measured
• Device counts per student rose while the cabling stayed put, so older blocks have one data point per room rather than one per desk

How many wireless devices does one classroom actually generate?

Count devices rather than people, and count the ones actually transmitting. A secondary student typically carries a laptop or tablet plus a phone; the phone is associated and mostly idle, yet it still costs airtime in beacons, keep-alives and management frames. A workable planning figure is two associated devices per student, with sixty to seventy-five percent active in a lesson.

Should a school run one unit per corridor or one per classroom?

Run both lines and take the larger. The coverage line is the one quoted by habit: one ceiling unit in the corridor serving two or three rooms. The capacity line divides concurrent devices by the 25 to 35 active clients one radio carries comfortably. Where the two disagree the capacity line decides, because a unit that reaches three rooms cannot also carry three rooms.

Take a 24-classroom secondary block with two devices per student and thirty students per room: about 42 concurrent active devices per room. One unit per classroom, with the two radios of a dual-band unit splitting the load, lands near 21 active clients per radio, inside the bracket. Planned as one corridor unit per three rooms, the same block puts roughly 126 devices on that unit, about 63 per radio, close to twice the design target.

Primary schools are where corridor placement survives: about 18 concurrent devices per room, three rooms per unit, roughly 27 active clients per radio. It stops working the day the school moves to one device per student, so placement follows the device plan rather than the floor plan.

Where should the access point physically go: corridor ceiling, classroom ceiling or a wall plate?

Placement changes the shape of the cell more than the model does. The table compares the three placements that actually show up in school projects.

What separates the models used in a classroom, a corridor and a dormitory?

Inside one classroom, two units with the same headline rate usually differ in the port behind them and the power they draw. A 3000Mbps unit on a gigabit port caps the room at roughly 940Mbps of real traffic in both directions; the same radio class behind a 2.5G port does not.

Every figure in the table comes from the product page of that model. Where a parameter is not published, such as the PoE class or the port layout on some panel units, ask for the datasheet for that model rather than inferring it.

What has to be switched on before the first lesson?

Placement follows the device plan, but half of the perceived result is configuration. The list is ordered by priority and needs no new hardware.

• Run 20MHz channels on 5GHz across the teaching block, and 20MHz on 2.4GHz as well
• Disable the 1, 2, 5.5 and 11Mbps rates, or the cell radius inflates and clients cling to the far end
• Set a minimum RSSI or kick threshold so the far seat moves to the nearer cell
• Enable 802.11k and 802.11v where the firmware supports them, and confirm per model rather than assuming
• Turn on airtime fairness, or one strong client can take a whole lesson
• Keep to two or three SSIDs: every SSID beacons on every channel
• Give every AP one uplink with headroom; a gigabit port carries about 940Mbps in both directions
• Match each unit consumption against the switch port PoE class before ordering

How do you accept a school WiFi installation before term starts?

Accepting a school network in an empty building repeats the survey mistake. Test with a class in the room, or load the room with simulated clients; the second is cheaper and repeatable.