A Beginner's Guide to Mobile Network Infrastructure

Phones feel like magic until the bars drop to one in a stairwell. Then it’s suddenly obvious that there’s actual hardware involved, sitting on a rooftop somewhere, connected by cables that someone had to dig a trench for.

Mobile networks aren’t that complicated once the pieces get laid out, and knowing them helps with practical stuff. Why an app crawls on a full 5G signal, say, or why websites treat phone traffic so differently from a laptop’s.

Three layers, and one of them is usually the problem

There’s the radio access network, the RAN, which is the wireless hop between a handset and the nearest tower. There’s the core, which checks the SIM, hands out addresses, and works out where packets should go. And then there’s transport: the fiber and microwave links tying it all together.

But transport gets ignored in the marketing and causes most of the disappointment. A tower can be bristling with 5G radios and still feel slow if its backhaul tops out at a few hundred megabits per second, which happens more often than carriers like to admit.

Towers are shared, too. One rooftop site often carries antennas from three or four operators, and the small cells bolted onto streetlights handle a surprising slice of city traffic.

The address situation is weirder than most people expect

A phone almost never gets its own public IP. Carrier-grade NAT stuffs hundreds of subscribers behind one routable address, which is why blocking a single cellular IP is such a blunt instrument. Ban it and a whole neighborhood goes with it.

So sites go easy on mobile ranges, which is exactly why those IPs are worth money to anyone doing ad verification or app testing at scale. Worth reading what is a mobile proxy at MarsProxies before signing anything, because the pricing model differs from datacenter IPs in ways that surprise people. The rules carriers follow for all this NAT machinery live in RFC 6888, down to how ports get allocated and sessions logged.

IPv6 is quietly killing off the whole arrangement. On several big US carriers most traffic already runs native IPv6, and every device gets a real address again.

2G to 5G: what actually changed

Speed is the headline every time, but the architecture underneath got rewritten with each generation. 2G digitized voice and threw in SMS almost by accident. 3G bolted packet data onto a design that still thought of itself as a phone system, and 4G LTE finally dropped circuit switching so voice became just another IP service.

5G’s headline feature is network slicing: one physical network split into several logical ones, each with its own latency and bandwidth guarantees. A factory running robots and a stadium full of people streaming can sit on the same towers without stepping on each other. The 3GPP 5G system specifications covering this run to thousands of pages, which is part of why rollout takes years.

And most 5G out there isn’t really 5G, structurally speaking. Non-standalone deployments run a 5G radio on top of a 4G core. Slicing and those sub-10ms latency numbers need a standalone core, which fewer operators have finished building.

What happens in the first second

Power on a handset and it scans broadcast channels, picks the loudest cell, and asks to join. The network checks the SIM against its subscriber database, issues temporary IDs, and sets up a data bearer. Usually done before the lock screen finishes animating.

The authentication trick is neat: a shared key sits on the SIM and in the carrier’s records, and it never crosses the air. Challenges get sent, responses come back, and the key itself stays put. That basic idea has survived since the 1990s with patches along the way.

Handover is the genuinely hard part. Cross a cell boundary at 70 mph and the network has to move a live session without losing packets, working off measurement reports the handset fires back dozens of times a second. The cellular network fundamentals behind that predate smartphones by decades.

Get it wrong and calls die in tunnels.

Where it’s going

Open RAN is the fight worth watching. Splitting radio hardware from the software running on it lets carriers mix suppliers instead of buying everything from Ericsson or Nokia, and operators in Japan and Germany have already put that to work in procurement talks.

Satellite direct-to-device is the other one, turning dead zones into a software problem instead of a construction project. Either way, the useful skill stays the same: figuring out which layer a problem lives on before spending an afternoon fixing the wrong thing.

Author’s Bio: I’m Harsh—Founder of saasit.in dedicated to helping SaaS and tech companies drive organic growth through strategic link building and data-driven outreach.

Author
Recent Posts

Brian Otieno

Brian Otieno is a technology writer covering broadband, mobile networks and consumer hardware. He reviews devices, untangles connectivity issues, and writes practical buying guides for everyday users. He also covers Kenya's phone market at Mzuri.