4G LTE Cellular Networks · Architecture and Elements

A 4G network is a radio edge and an all- IP (Internet Protocol) core, built by an industry that had never used IP. It arrived at most of the same architectural ideas as the Internet anyway.

Words you will meet

Cell — one geographic coverage area, with a base station in it.
Base station — officially “eNode-B”. Manages the radio, and rather more.
MME (Mobility Management Entity) — authenticates devices and sets up their paths. Never carries data.
HSS (Home Subscriber Service) — the database of a network’s own subscribers.
S-GW (Serving Gateway) and P-GW (PDN Gateway) — the two routers on the data path. The second does network address translation .
IMSI — the subscriber’s globally unique identity, stored on the SIM card.

Why this matters

WiFi (IEEE 802.11 wireless LAN) coverage is patchy, and you cannot associate with every access point you walk past. 4G is not. A study of over a million US subscribers found a 4G signal more than 90 per cent of the time; Korean carriers manage between 95 and 99.5 per cent.

But the interesting part is not the coverage. Cellular networks came out of the telephony world, whose architecture had almost nothing in common with the Internet’s. And yet the book’s own framing is that you will recognise nearly everything here.

Protocol layering. An edge and a core. A network of networks. And a clean separation of data and control planes.

Two industries, starting from different places, arriving at the same architecture. That is worth more than any single mechanism on this page.

The names are bad, and the book says so

Read this before the diagrams

The book warns that 4G LTE (Long-Term Evolution) is “full of rather obtuse acronyms and element names”. Then it does something unusual: it replaces two of them with plain words for the rest of the chapter. “User Equipment” becomes mobile device; “eNode-B” becomes base station. This page does the same.

Its etymology of eNode-B is worth reading once. 3G called network function points nodes. The B goes back to 1G’s Base Station and 2G’s Base Transceiver Station. And since 4G is an evolution of 3G, an e was added to the front. The book then notes that 5G calls the same function ng-eNB and invites you to guess what that stands for.

Do not let the names do the work. Every element here does something simple, and the table below says what.

The acronyms, decoded once

Official name	What this page calls it	Its job

Cells marked ⓘ have an explanation — click to read it.

The book itself apologises for these names, and uses reader-friendly substitutes for two of them. This page does the same. Click a row for what each element actually does.

Two planes

Figure 7.18 — the elements, split into two planes

4 hops: mobile device → base station → S-GW → P-GW → Internet

Drag any device to rearrange the picture. Hover a link to see its rate, delay and length.

Switch between the journeys and watch which elements each one touches. The MME authenticates you and builds your path, and then never sees a single one of your datagrams. That is chapter 5’s data plane / control plane split, arrived at independently by the telephony world.

Read this diagram as text

base station — wired link — MME — authenticates, builds tunnels
MME — authenticates, builds tunnels — wired link — HSS — the subscriber database
MME — authenticates, builds tunnels — wired link — S-GW
S-GW — wired link — P-GW
mobile device — wireless link — base station
base station — wired link — S-GW
S-GW — wired link — P-GW
P-GW — wired link — Internet

In plain words

Switch between the two journeys and watch which elements each one touches.

The MME authenticates you, arranges your path and remembers which cell you are in. And then it never sees a single one of your datagrams — the book is explicit that it is not in the forwarding path.

That is chapter 5’s separation of the data plane from the control plane, with logically centralised control. The telephony industry arrived at it independently.

Everything here, against something you already know

Every 4G element, against something you already know

	4G LTEbuilt by the telephony world	The Internetchapters 1–6
The thing at the edge
What it attaches to
Its permanent identity
Who decides how traffic flows
Where the network meets the outside
How the global system is organised

Cells marked ⓘ have a reason behind them — click to read it.

The book says outright that 4G reuses architectural principles from chapters 1 to 6. This is that sentence, made checkable. Click any cell — several of the differences matter more than the similarities.

Two rows repay attention, and both are places where the correspondence is imperfect:

A base station is not an access point. It manages radio and channel access much as an AP does. But it also builds device-specific IP tunnels, coordinates handovers with neighbouring base stations, and negotiates spectrum with them to limit interference between cells. The book’s phrase is “comparable (but by no means identical)”, and three extra jobs is a wide gap.

An IMSI is not quite a MAC (Media Access Control) address. Both are flat, permanent and globally unique, and the book says the analogy holds “in some ways”. But a MAC address names an adapter, and an IMSI names a subscriber — which is why you can move it to another phone.

Attaching

Attaching to the network

The device asks its base station to attachstep 1 of 5

The mobile device arrives in a cell and asks to join. Everything from here until it has an address is control plane — no user data moves at all.

Five steps, and none of them carries user data. The whole of this happens before the first datagram — and the MME, which arranged all of it, then drops out of the picture entirely.

Read all steps as text

The device asks its base station to attach — The mobile device arrives in a cell and asks to join. Everything from here until it has an address is control plane — no user data moves at all.
The local MME asks the home network — The MME contacts the HSS in the device’s HOME network — which may belong to a completely different carrier, in a different country, if the device is roaming.
Both sides prove themselves to each other — The home HSS returns enough encrypted information for the device to be sure the network is a legitimate carrier, and for the network to be sure the device really is the one holding that IMSI. Mutual authentication — chapter 8 covers how.
The MME builds the path — as tunnels — A wireless first hop, then an IP tunnel from base station to S-GW, then another from S-GW to P-GW. Concatenated tunnels, not ordinary router-by-router forwarding.
And the P-GW gives it an address — A NAT address, from the P-GW. From here the device is an ordinary Internet host — and everything about its mobility is hidden behind that gateway.

Why tunnels, rather than ordinary forwarding?

The data path is not built out of routers forwarding hop by hop. It is a wireless first hop and then two concatenated IP tunnels — base station to S-GW, and S-GW to P-GW — built under the MME’s control.

The reason is mobility, and it is the whole point of the design. When a device moves, only a tunnel endpoint has to change. Nothing else on the path needs to learn anything, and no routing table anywhere has to converge.

Section 7.6 is where that pays off.

The P-GW hands the device a NAT (Network Address Translation) address and performs the translation, exactly as section 4.3.4’s home gateway does — at a rather different scale. To everything beyond it, the P-GW is just another gateway router, and every bit of mobility inside the carrier’s network is invisible.

Everyday picture

Checking into a hotel abroad. The front desk cannot vouch for you itself, so it phones your bank at home. The bank sends back enough to satisfy both sides: the hotel that you are who you claim, and you that the hotel is a real hotel.

Then the desk gives you a room and a key, and takes no further part. It does not follow you around, and it never sees what you do.

Where the picture stops. A hotel room stays where it is. The 4G equivalent is a path built out of tunnels precisely so that the room can be moved without telling anybody — which is section 7.6.

Check yourself

0 of 7 answered

1.Does the MME carry a mobile device's datagrams to and from the Internet?
2.The book says a cellular base station is "comparable (but by no means identical)" to a WiFi access point. What does it do that an AP does not?
3.predictA device roaming on another carrier's network tries to attach. Which HSS is consulted?
4.Why is the data path built from concatenated IP tunnels rather than ordinary hop-by-hop forwarding?
5.What does the P-GW do, and what does the outside world see of it?
6.The book says an IMSI is "in some ways analogous" to a MAC address. Where does the analogy stop?
7.The book says 4G reuses architectural principles from chapters 1 to 6. Which of these is NOT one it names?

What to remember

A 4G network is a radio edge and an all-IP core. The MME authenticates and builds paths, and is never in the forwarding path — chapter 5’s control plane, reinvented.
The P-GW is the last LTE element before the Internet: it hands out addresses, does NAT, and hides all mobility behind itself.
The data path is concatenated IP tunnels, not hop-by-hop forwarding, so that a move changes only an endpoint. And an IMSI names a subscriber, not an adapter, which is where the MAC-address analogy stops.