JOVANA
Explore Library Glossary Getting Started Three Levels Fields How it works Mission
Join the mission
All guides

Cellular Networks: Cells, Base Stations, and the Core

Wi-Fi gives you a room; the cellular network gives you a continent. Here is how engineers cover a whole country with radio by slicing the land into cells, and how the hidden 'core' behind the towers quietly turns your phone into just another thing on the Internet.

Why not one giant transmitter?

In the last guide you saw how Wi-Fi covers a room or a building: one access point, a shared slice of radio, and everyone politely taking turns. The obvious way to cover a whole city would be to make that idea bigger — one enormous antenna on the tallest hill, blasting at full power so every phone for fifty kilometres can hear it. That instinct is exactly wrong, and seeing why is the key to the entire cellular idea. Radio spectrum is a fixed, scarce resource: there are only so many frequencies, and a regulator hands out narrow bands at auction for billions. One giant transmitter would have to share that one band among every phone in the city at once — millions of people fighting over the airtime that CSMA taught you is always limited.

The cellular insight flips the problem. Instead of one big cell, carve the land into many small cells, each covered by a modest tower of its own. The magic is that a frequency used in one cell can be reused in another cell far enough away that the two signals do not reach each other — remember attenuation from the first guide in this rung, the way a radio signal fades fast with distance. That fading, which felt like a problem, is now a gift: it lets the same precious frequencies be spent over and over across the map. This is frequency reuse, and it is why the system is called cellular — the coverage map looks like a honeycomb of cells, traditionally drawn as hexagons because hexagons tile a plane with no gaps.

The base station: the tower that is really a switch

At the centre of every cell sits a base station — the radio tower you see by the highway, though most of the cleverness is in the equipment cabinet at its base, not the metal. Its job has two faces. Facing the phones, it is a radio: it owns the cell's frequencies and decides, moment to moment, which phone gets to transmit and when. Unlike a Wi-Fi access point, where devices grab airtime by listening and avoiding collisions, the base station is a tight scheduler — it tells each phone exactly which slice of frequency and which slot of time to use. This is why a crowded cell still feels orderly: nobody is shouting over anyone, because a referee is assigning turns.

Facing inward, the base station is the on-ramp to a wired network. It is not the end of the journey — it is the first hop. Behind it runs a backhaul link, usually fibre, carrying the traffic of thousands of phones back toward the operator's machinery. So a base station is best understood as a special kind of switch: on one side a carefully scheduled radio, on the other a fat fibre into the wired world. When you send a packet from your phone, it crosses the air to the nearest base station, and from there it travels over ordinary cables exactly like any other packet you have studied. The wireless part — the hard, lossy wireless link you met in the first guide of this rung — is just the very first and very last hop.

The core: the brain behind the towers

Behind all the base stations sits the part travellers never see: the core network, or just 'the core'. If the base stations are the doors of the building, the core is everything inside — the registry that knows every subscriber, the gatekeeper that checks you are a paying customer, the bookkeeper that counts your data, and the gateway that finally connects you to the rest of the Internet. The towers are dumb and replaceable; the core is where a phone network's real identity and intelligence live. When people say a network is '4G' or '5G', they are talking as much about the design of this core as about the radio.

The core has to solve a problem the wired Internet never faced: subscribers move and must be found. Every phone has a permanent identity stored on the SIM card, separate from any IP address. When you power on, the core authenticates you against that identity, and — crucially — it remembers which base station you are currently near, in a kind of live address book. So the core does two jobs at once that the ordinary Internet splits across many systems: it is the address-and-identity desk that knows who you are, and it is the location service that knows where you are right now. Only after the core has placed you does it hand you an IP address and let your packets out onto the global Internet.

  YOUR PHONE                                              THE INTERNET
      |                                                        ^
      |  (1) radio link: scheduled,                            |
      |      lossy, scarce spectrum                            |
      v                                                        |
  [ BASE STATION ] --(2) backhaul fibre--> [ CORE NETWORK ] ---+
   one per cell;                            - who are you? (SIM)
   schedules the air;                       - where are you? (which cell)
   first wired hop                          - count the data, charge it
                                            - gateway out to the Internet
The shape of every cellular network: a scheduled radio hop to a base station, a wired backhaul to the core, and the core acting as identity desk, location tracker, meter, and gateway. The Internet only sees you after the core lets you through.

Moving without dropping: handoff

Now the feature that makes a phone a phone: you can drive at highway speed through dozens of cells without your call or download ever pausing. As you leave one cell and enter the next, the network must move your connection from the old base station to the new one — a handoff (also called handover). Picture a relay race where the baton is your live session: the runner must pass it cleanly while both are still running, because if it drops, your call cuts out. Your phone is constantly measuring the signal strength of nearby cells and reporting it back, so the network can see a stronger neighbour coming before the current one fades.

  1. Your phone, while still talking to base station A, measures that neighbouring base station B is now stronger and reports this back over the current link.
  2. The core network tells base station B to prepare resources for you — a frequency and time slot in its cell are reserved before you arrive.
  3. At the chosen instant your phone retunes its radio to B and the core re-points your traffic so packets now flow through B instead of A.
  4. Base station A releases the resources it was holding for you, freeing that airtime for someone else in the old cell.

Done well, a handoff takes a fraction of a second and you never notice. But notice what it costs: the network is doing real work — measuring, signalling, reserving, switching — just to keep your single connection alive while you move. This is mobility management, and it is the deep reason a cellular core is so much more complicated than a wired router. A router only has to answer 'where is this address?'; a cellular core has to keep answering 'where is this person, right now, this second?' — and the answer changes every time you turn a corner.

Why moving and wireless loss confuse TCP

Here is where a beautiful idea from earlier in the ladder runs into trouble. Recall TCP's congestion control: the careful driver who speeds up gently and brakes hard, reading every lost packet as a sign that the network is jammed and immediately slowing down. That assumption — loss means congestion — was true on wired links, where bits almost never get corrupted in flight. On a wireless link it is often false. A packet can be lost because a wall got between you and the tower, or interference flared, or the bit error rate spiked for a moment — not because anything is congested at all.

So classic TCP misfires: it sees a wireless loss, assumes congestion, and slams on the brakes — cutting its sending rate hard even though the path was never full. The link was simply briefly noisy, and the right move would have been to resend that one packet and keep going. Instead, throughput collapses for no good reason. This is the heart of the wireless TCP performance problem, and it is an honest, well-known wart, not a detail we are smoothing over. It is the price of a layered design: TCP, sitting high up, cannot tell why a packet vanished, so it falls back on the only assumption it was born with.

Putting it together

Step back and the whole architecture is a single clean idea repeated at scale. Carve space into cells so scarce spectrum can be reused; put a scheduling base station at the heart of each cell as both a radio and the first wired hop; hang all the brains in a core network that knows who and where every subscriber is, meters their data, and bridges them to the global Internet. Layered on top, mobility management keeps your one connection alive through handoff after handoff as you move. And honestly threaded through all of it is the tension with TCP, which was never designed for a path that loses packets to physics rather than congestion.

One honest caveat to close on: 'cellular' is a family of generations, not one fixed design, and the boundary between 4G LTE and 5G is partly real engineering and partly marketing. The bones in this guide — cells, base stations, a core, handoff — are common to all of them. What changes between generations is how fast the radio runs, how flat and software-driven the core becomes, and how cleverly each layer hides wireless loss from TCP. That is exactly the story of the next guide.