Skip to content
Tutorials Center
My learning

Topics

Lesson 60 of 79 ยท Wireless and Mobile Networks

Cellular Networks: 4G, 5G and Beyond

How mobile networks work: cells and frequency reuse, handoff, the radio access network and core, LTE architecture, 5G features such as slicing and mmWave, and IoT.

A cellular network lets millions of people move around while their calls, video streams and web sessions keep running. Unlike Wi-Fi, which covers a room or a building from one cheap box, a cellular network covers whole countries with licensed spectrum, carrier-grade equipment and a core network that tracks every subscriber.

The two ideas that make it work are old and simple: divide the territory into cells and reuse the same radio frequencies in distant cells, and hand a moving user from one cell to the next without dropping the connection. This lesson explains those ideas, the split between the radio access network and the core, how 4G (LTE) is built, and what 5G added. Nothing here can run on a laptop, so the examples are small honest models of the physics and the decisions involved, not a real radio stack.

Cells and frequency reuse

Radio spectrum is scarce. A single powerful tower covering a whole city could serve only as many calls as it had channels. The cellular idea is to use many low-power base stations, each covering a small area (a cell, from a few hundred meters in a city to tens of kilometers in the countryside). Because signals fade with distance, two cells far enough apart can use the same frequencies without disturbing each other. This is frequency reuse.

Classic designs group cells into clusters of N cells, where every cell in the cluster gets a different set of frequencies and the pattern repeats. N must have the form i*i + i*j + j*j for non-negative integers i and j, so the common sizes are 1, 3, 4, 7, 9 and 12. A larger cluster separates co-channel cells more and reduces interference, but each cell gets only 1/N of the channels. A reuse factor of 7 was a typical analog-era choice. Modern LTE and 5G networks often use reuse 1 (every cell uses the whole band) and manage the interference at the edges with scheduling and coordination.

To add capacity in a busy area, operators split cells (smaller cells, more towers), add sectors (a tower with three directional antennas serves three cells), or add small cells in stadiums and city centers.

Python Cluster sizes and capacity

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Handoff

When a user drives away from a tower, the signal weakens and a neighbor's signal strengthens. The network must move the connection to the new cell, called handoff (or handover), without the user noticing.

  • Hard handoff ("break before make"): the phone releases the old cell and then connects to the new one. LTE does this, with very short gaps (tens of milliseconds) and data forwarded between towers so little is lost.
  • Soft handoff ("make before break"): the phone is connected to two cells at once for a while. Older CDMA/3G networks used it.

The decision uses measurements: the phone constantly reports the signal strength of neighbor cells, and the network (not the phone, unlike Wi-Fi) decides. To avoid the ping-pong effect, where a user standing at the border bounces between two cells, networks add a hysteresis margin (the neighbor must be better by some dB) and a time-to-trigger (it must stay better for some time).

sequenceDiagram
    participant UE as Phone
    participant S as Serving tower
    participant T as Target tower
    UE->>S: Measurement report (target is stronger)
    S->>T: Handover request
    T->>S: Handover accepted
    S->>UE: Handover command
    UE->>T: Connect to new cell
    S-->>T: Forward buffered data
    T->>UE: Data continues

A simplified handover between two towers. The user's IP address does not change.

Python Handoff with hysteresis and time-to-trigger

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The generations from 1G to 5G

Mobile networks are labeled by generation. A new generation roughly every ten years brought a new radio technology and usually a new kind of service. The table summarizes them; speeds are typical figures, not guarantees.

Generation Era Technology What it delivered
1G 1980s Analog voice, FDMA (for example AMPS) Mobile phone calls only. No encryption, so calls could be overheard, poor capacity and big handsets
2G 1990s Digital, TDMA (GSM) and CDMA Digital voice with encryption, SMS text messages, and later packet data with GPRS and EDGE (tens to a few hundred kbit/s)
3G 2000s WCDMA/UMTS, CDMA2000, later HSPA Mobile Internet, video calls and the first smartphones (hundreds of kbit/s up to tens of Mbit/s with HSPA)
4G 2010s LTE and LTE-Advanced, OFDMA, MIMO An all-IP network with voice as VoLTE, HD video streaming and apps (tens to hundreds of Mbit/s)
5G 2020s 5G NR, wide channels, massive MIMO, beamforming, slicing Faster speeds, very low latency, huge numbers of IoT devices

Three changes stand out. First, the move from analog to digital (2G) allowed encryption and efficient use of spectrum. Second, the move from circuit-switched voice to packet data (2.5G/3G) put the Internet on the phone. Third, with 4G the whole core became IP, so a voice call is just another data flow. 5G is designed around three service types: eMBB (enhanced mobile broadband), URLLC (ultra-reliable low-latency communication, for factories and remote surgery) and mMTC (massive machine-type communication, for sensor networks).

Python How long does a 2 GB download take in each generation?

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The numbers are ideal averages, but they show why each generation opened new uses: a movie download that took days on 2G finishes in seconds on good 5G.

The radio access network and the core

A cellular network has two big parts:

  • The RAN (Radio Access Network) is the towers and antennas. It handles the radio link: scheduling who transmits when, modulation, error correction and handoff. In 4G the tower is the eNodeB; in 5G it is the gNB.
  • The core network is the carrier's own IP network behind the towers. It authenticates subscribers (using the SIM card), tracks where each phone is, assigns IP addresses, enforces plans and billing, and connects to the Internet and to other carriers. Voice, SMS and data are all carried over it.

LTE (4G) architecture

LTE flattened and simplified the older networks into an all-IP design called the EPC (Evolved Packet Core). The main parts:

Component Role
UE User equipment: the phone, with its SIM
eNodeB The base station; radio scheduling, handover decisions
MME Mobility Management Entity: authentication, tracking which area the phone is in, paging
HSS The subscriber database (keys, plan, allowed services)
S-GW Serving gateway: anchor point for user traffic while the phone moves between towers
P-GW Packet gateway: assigns the phone's IP address and connects to the Internet

The design separates the control plane (signalling: attach, authenticate, move) from the user plane (your actual packets). A phone, when it attaches, gets a bearer: a tunnel (using GTP over UDP) from the phone to the P-GW with a defined quality of service. Your IP address stays the same as you move because the tunnel simply gets re-anchored at a different tower. The P-GW is where NAT, usually carrier-grade NAT, typically sits, so a phone's public address is shared with many other subscribers.

flowchart LR
    UE["Phone (UE)"] --- ENB["eNodeB"]
    ENB -->|"user plane"| SGW["S-GW"] --> PGW["P-GW"] --> NET["Internet"]
    ENB -.->|"control plane"| MME["MME"]
    MME --- HSS["HSS"]
    MME -.-> SGW

Solid lines carry user packets; dotted lines carry signalling.

Mobility is the big difference from Wi-Fi at the network layer. In Wi-Fi roaming the AP change is invisible because it happens within one subnet. In cellular, the phone may be hundreds of kilometers from where it started, and the core hides that by tunneling to a fixed gateway.

5G: what changed

5G is a family of new radio technologies (NR, New Radio) and a new core. The marketing focuses on speed, but the engineering goals are three different service types:

  • eMBB (enhanced mobile broadband): much higher throughput, for video and crowded places.
  • URLLC (ultra-reliable low-latency communication): targets of a few milliseconds and very high reliability, for industrial control and remote machinery.
  • mMTC (massive machine-type communication): a very large number of cheap, low-power sensors per square kilometer.

Three spectrum ranges

Range Example frequencies Behavior
Low band below 1 GHz Excellent range and building penetration, modest speed gain over 4G
Mid band 1 to 7 GHz (3.5 GHz is common) The sweet spot: wide channels, still reasonable range
High band (mmWave) 24 to 52 GHz and above Huge bandwidth and multi-gigabit speeds, but very short range and easily blocked by walls, rain and even a hand

The physics is free-space path loss: for the same distance, loss grows with the square of the frequency, so 28 GHz loses about 32 dB more than 700 MHz over the same distance (20 times log10 of 40). mmWave therefore needs dense small cells, and it relies on massive MIMO and beamforming: antenna arrays with dozens or hundreds of elements that steer a narrow beam toward each user instead of radiating in all directions.

Python Why mmWave needs small cells

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Network slicing and the service-based core

The 5G core is built from software network functions that talk to each other with web-style APIs (HTTP/2 and JSON) instead of dedicated hardware boxes with custom protocols. The main ones: the AMF handles access and mobility (similar to the MME), the SMF manages sessions, the UPF forwards user traffic (like the S-GW/P-GW) and can be placed close to users, and the AUSF/UDM authenticate subscribers.

Because these are software, a carrier can run several virtual networks on the same physical infrastructure: network slicing. One slice might offer low latency and guaranteed capacity for a factory, another cheap best-effort bandwidth for phones, a third high reliability for emergency services. Each slice has its own core functions, policies and quality-of-service rules.

Edge computing (MEC) goes with this: placing servers and UPFs near the towers cuts the round-trip time for applications such as cloud gaming, AR or machine vision.

Many 5G deployments began as non-standalone (NSA): 5G radio attached to the existing 4G core. Standalone (SA) networks use the new 5G core and unlock slicing and the lowest latencies.

Cellular IoT

Millions of meters, trackers and sensors need coverage but not speed, and they must run for years on a battery. Cellular standards include dedicated low-power modes:

  • LTE-M (Cat-M1): up to about 1 Mbit/s, supports mobility and voice, moderate power.
  • NB-IoT (Narrowband IoT): a very narrow channel (about 200 kHz), tens of kbit/s, excellent building penetration and battery life of ten years or more.

These use power-saving modes and extended sleep cycles: the device wakes briefly, sends a small reading, and sleeps again. The next lesson compares them with other short- and long-range IoT radios in Bluetooth, Zigbee, LoRa and IoT networking.

Beyond 5G

Research on 6G aims at frequencies above 100 GHz, sensing built into the radio (the network doubles as a radar), tighter integration with AI for scheduling, and satellite links. Satellite-to-phone service is already reaching real products, with non-terrestrial networks (NTN) being standardized as part of 5G. Standards bodies such as 3GPP release a new set of features roughly every 18 months, so the details keep moving, but the core concepts here (cells, reuse, handover, control versus user plane, tunneling to a gateway) stay the same.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. What makes frequency reuse possible in a cellular network?

    Radio signals fade as they travel. Cells that are far enough apart can reuse the same channels without causing serious interference.
  2. 2. Why do networks add hysteresis and a time-to-trigger to handover decisions?

    Without a margin and a delay, a phone at the border between two cells could bounce back and forth every time the measured signal crossed over.
  3. 3. In an LTE network, which component assigns the phone's IP address and connects it to the Internet?

    The packet gateway is the exit point of the core network. The phone's traffic is tunneled to it, which is why the address stays the same while the phone moves.
  4. 4. Why does mmWave 5G need many small cells?

    Path loss grows with frequency and obstacles absorb mmWave signals, so each cell covers only a short distance. The benefit is a very large amount of bandwidth.
  5. 5. What is network slicing?

    Slicing relies on the software-based 5G core. Each slice can have its own functions and quality-of-service rules, for example one for a factory and one for ordinary phones.

Summary

  • Cellular networks divide an area into cells and reuse frequencies in distant cells; capacity grows by splitting cells, adding sectors and small cells.
  • Handoff moves a connection between towers; hysteresis and time-to-trigger prevent ping-pong. In cellular, the network decides.
  • The RAN handles the radio; the core authenticates, tracks and routes. LTE's EPC has eNodeB, MME, HSS, S-GW and P-GW, with user-plane traffic tunneled to a fixed gateway.
  • 5G targets eMBB, URLLC and mMTC, uses low, mid and mmWave spectrum, beamforming, a software service-based core and network slicing.
  • NB-IoT and LTE-M serve low-power devices on the same cellular infrastructure.

Next, look at short-range and low-power radios: Bluetooth, Zigbee, LoRa and IoT networking.