Most campus network advice is written from the outside. This is written from the inside. We analysed data from a live network at a leading Indian university which has 90+ colleges, 86 academic departments, 130,000+ students. The analysis resulted in valuable insights on how 

1. users interact with the network
2. load varies as a function of time
3. mobility impacts the network
4. user density and data usage are not always in-sync

A Campus Network That Runs Around the Clock

The first lesson is the simplest and the most overlooked: there is no "off" on a campus. Across the week, daily traffic and client counts stayed high through the working days, as expected. Even on a public holiday, and the network carried substantial load. Students in hostels, researchers, automated systems, and background services do not stop because the calendar says so.

This matters for design. A network you can take down for maintenance on a quiet evening does not really exist here. High availability is not a premium feature on a campus; it is the baseline, because there is no window when the network is genuinely idle.

The Peak Hour That Shapes Capacity Planning

Traffic on an education campus is not spread evenly through the day. It concentrates, sharply. In this deployment, 12 to 1 PM was consistently the busiest hour of every day. During that single hour, on average almost 50% of the entire day's unique clients were active on the network. 

Think about what that means for sizing. If you design for the daily average, you have massively under-built for the hour that actually matters. 50% of the target userbase is online in the same 60-minute window, between classes, over lunch, and the network must hold up precisely when the most people are watching. Peak-hour capacity, not average capacity, is the number that defines whether your campus network succeeds or fails.

The AP-Load Paradox: The Difference Between Clients and Capacity

Here is the finding that surprises most people, and it has direct consequences for how you plan. The access points with the most clients may not be the access points carrying the most traffic.

Ranked by client count, the top access points were in colleges and common areas, one served 857 unique clients in a day, another 775. However, when ranked by data traffic, the top access points were in classrooms and libraries. One classroom access point carried 156 GB in a day while serving only 240 unique clients whereas a library access point carried 119 GB from 541 unique clients.

The pattern is clear: classrooms and libraries are the real hotbeds of activity. They combine focused, data-heavy use, streaming, downloads, research, uploads, in one place. A crowded corridor access point with 800 casual clients checking phones moves far less data than a classroom of 240 people all streaming a lecture. If you plan capacity by headcount alone, you will over-build the corridors and under-build the classrooms.

Most Access Points Are Ideal. The Campus Traffic Distribution

Zoom out to the whole network and the load is strikingly uneven. Two distributions from the week tell the story.

By client count: 95% of access points served fewer than 250 unique clients in a 24-hour window. The vast majority carry a light load. Only about 1% saw 500 clients in a day.

By data traffic: roughly 35% of access points carried the bulk of the traffic, and about 5% each moved more than 10 GB in a single day. A small minority of access points do most of the heavy lifting.

The design lesson is precise: don’t deploy based on per square foot rule. A uniform network, identical access points everywhere, wastes money on the quiet zones and starves the hotspots. This is exactly why zone-based design (matching high-capacity access points to classrooms and auditoriums, lighter ones to corridors and hostels) beats a one-size-fits-all rollout.

Two Out of Three Clients Are on 5 GHz - But 2.4 GHz Still Matters

Across the week, roughly two in three clients connected on the 5 GHz radio. That is the healthy, expected pattern: 5 GHz offers more capacity and less congestion, and modern devices prefer it.

But the remaining third on 2.4 GHz is not noise. That band carries a real, non-trivial share of clients, older devices, IoT endpoints, and devices at the edge of coverage. It cannot be switched off. The practical implication, which we will return to in the companion piece on deployment lessons, is that 2.4 GHz has to be deployed deliberately and carefully, not ignored, because a meaningful slice of the campus depends on it.

Mobility Is Bimodal: Most Stay Put, Some Roam Constantly

How much do campus users actually move between access points in a day? The data shows a clear split of users into two groups.

44% of clients showed limited mobility, associating with fewer than five access points in a 24-hour window. These are people who settle: in a lab, an office, a hostel room, a favourite library corner.

At the other end, about 10% of clients connected to 25 or more different access points in a single day. These are the true roamers, moving across the campus continuously.

That 10% is small in number but heavy in impact, because every move triggers a fresh association, a DHCP request, and an authentication handshake. A network that handles the sedentary 44% comfortably can still be brought to its knees by the roaming 10% if the systems behind roaming (DHCP and AAA) are under-sized. Mobility is a design input, not an afterthought.

Uplink Traffic Is Bigger Than You Think

The prevailing industry consensus is that networks are download-heavy: users pull content down, send little up. On a modern campus, that is breaking. In this deployment, more than 30% of access points had significant uplink traffic, and about 10% actually carried more upload than download.

The drivers are obvious once you look: video conferencing, cloud backups, shared drives, online submissions, research data moving out to the cloud. This changes how you provision. A network designed with a heavily asymmetric, download-first assumption will choke on the upstream. Uplink capacity has to be planned for, not treated as an afterthought.

The Data Snapshot

One ordinary week, one real campus, in numbers:

• ~18.5 TB of traffic in a single day
• ~50,000 unique clients on a working day
• 7,500+ live access points
• 12 to 1 PM the busiest hour, every day
• ~50% of the day's clients active in that one hour
• 95% of APs served under 250 clients; 5% carried over 10 GB each
• 857 clients on the busiest AP; the heaviest-traffic APs were classrooms and libraries
• 2 in 3 clients on 5 GHz; the rest on 2.4 GHz
• 44% of clients barely moved; 10% roamed across 25+ APs in a day
• 30%+ of APs carried significant uplink traffic

None of this is theoretical. It is what a 130,000-student network actually did in one regular working week. The obvious next question is what to do about it: how to plan RF, size DHCP and AAA, set PoE budgets, and design SSIDs so a network handles load like this. I have written those field lessons up separately, in Scaling Campus Wi-Fi: 10 Hard-Won Lessons from a 130,000-Student Deployment, the practical companion to this data.

This analysis is drawn from real operational data on a live io by HFCL campus network. io by HFCL designs, manufactures, deploys, and manages smart campus networks end-to-end, from Wi-Fi access points and switches to the io Canvas AI management platform. To understand what your own campus traffic looks like, request a free network assessment.