- Advanced
- About 16 min
- 5 practice questions
5G in depth: spectrum trade-offs, massive MIMO, network slicing and edge computing
Reason about 5G spectrum bands from wavelength, explain massive MIMO, beamforming, network slicing and standalone 5G, calculate latency gains from edge computing, and assess IoT security.
If something looks wrong, please tell us.
At the advanced level, questions on 5G ask why rather than what: why millimetre-wave signals need many small cells, why a factory and a video app can share one network without disturbing each other, why moving a server closer cuts delay. Each answer comes from a little physics and a few design ideas. This lesson builds them and ends with the security weakness that comes with billions of connected devices.
Spectrum: the coverage-capacity trade-off
| Band | Rough range | Strength | Weakness |
|---|---|---|---|
| Low band | Below 1 GHz | Wide coverage; passes through walls well | Limited capacity and speed |
| Mid band | About 1 to 6 GHz | Balance of coverage and capacity | Needs more towers than low band |
| High band (millimetre wave) | About 24 GHz and above | Very high capacity and speed | Short range; blocked by walls, foliage and even the body; needs many small cells |
Worked example 1: wavelength explains the trade-off
Wavelength = speed of light ÷ frequency, with light at 3 × 10^8 m/s. At 700 MHz: 3 × 10^8 ÷ 7 × 10^8 ≈ 0.43 m, or 43 cm. At 28 GHz: 3 × 10^8 ÷ 2.8 × 10^10 ≈ 0.0107 m, or about 1.07 cm. Waves only about a centimetre long are absorbed and blocked far more easily, so millimetre-wave cells must be small and close together. The same short wavelength has a benefit: antenna elements scale with wavelength, so dozens or hundreds of tiny antennas fit into one panel.
Technologies that make 5G work
- Massive MIMO (multiple input, multiple output) - a base station uses a large array of antennas to serve many users at once on the same frequency.
- Beamforming - instead of spraying signal in all directions, the array focuses a narrow beam towards each user, increasing range and reducing interference.
- Small cells - low-power base stations on poles and buildings that add capacity, essential for millimetre wave.
- Software-defined networking (SDN) and network function virtualisation (NFV) - network functions run as software on general-purpose servers instead of dedicated boxes, so the network can be reconfigured quickly.
- Network slicing - one physical network is divided into several virtual networks, each tuned for a purpose: one slice for high-speed video, another for ultra-reliable machine control, another for millions of low-power sensors.
- Multi-access edge computing - small data centres at the edge of the operator's network run applications close to users.
Non-standalone (NSA) 5G uses 5G radios anchored to an existing 4G core network, allowing a quick start. Standalone (SA) 5G uses a new 5G core and is needed for the full feature set, including network slicing. The ITU's minimum technical requirements for IMT-2020 include a peak download rate of 20 Gbit/s, user-plane latency of 1 ms for low-latency services and a connection density of one million devices per square kilometre - design targets, not the speed a user should expect.
Worked example 2: how much does edge computing save?
Light travels through optical fibre at about two-thirds of its speed in vacuum, about 200,000 km/s. A server 1,500 km away: one-way delay = 1,500 ÷ 200,000 = 0.0075 s = 7.5 ms, so a round trip is 15 ms before any processing. An edge server 15 km away: one way = 15 ÷ 200,000 s = 0.075 ms, round trip 0.15 ms. Distance alone costs a hundred times more delay for the distant server. That is why low-latency services need computing at the edge: no radio upgrade can beat the speed of light.
Higher frequency → shorter wavelength → more capacity, less reach, smaller antennas. Slicing = many virtual networks on one physical network. Round-trip delay in fibre ≈ 2 × distance ÷ 200,000 km/s.
IoT security
Many IoT devices are cheap, rarely updated and shipped with default passwords. In 2016 the Mirai botnet took over large numbers of insecure cameras and routers and used them for massive distributed denial-of-service (DDoS) attacks that disrupted major websites. Good practice: unique passwords per device, security updates for the device's life, encryption of data in transit, collecting only the data needed, and separating IoT devices from critical networks.
Key takeaways: wavelength explains why low bands cover and high bands carry. Massive MIMO and beamforming reuse spectrum; slicing and NFV make one network serve very different needs; full slicing needs standalone 5G. Physical distance sets a floor on delay, which edge computing lowers. Billions of weakly protected devices are a security problem in their own right.
Check your understanding
-
1. What is the approximate wavelength of a 30 GHz radio signal (speed of light 3 × 10^8 m/s)?
- 10 cm
- 1 mm
- 1 cm
- 10 m
Show answer
Answer: C. Wavelength = 3 × 10^8 ÷ 3 × 10^10 = 0.01 m = 1 cm. This is why such bands are called millimetre-wave or centimetre-wave and are easily blocked.
-
2. An operator runs a high-speed video service and an ultra-reliable control service for a factory on the same physical 5G network, each with its own guaranteed performance. The feature that makes this possible is:
- Network slicing
- Analog modulation
- Proof of stake
- SMS gateways
Show answer
Answer: A. Network slicing divides one physical network into virtual networks, each configured for a particular kind of service.
-
3. Consider the statements: (1) Non-standalone 5G uses 5G radios with an existing 4G core network. (2) Millimetre-wave 5G cells cover larger areas than low-band cells. Which is/are correct?
- 2 only
- Both 1 and 2
- Neither 1 nor 2
- 1 only
Show answer
Answer: D. Statement 1 is correct: NSA anchors 5G radios to the 4G core. Statement 2 is wrong: millimetre waves have short range, so their cells are smaller than low-band cells.
-
4. A server is 600 km away by optical fibre, where signals travel at about 200,000 km/s. Ignoring processing, the round-trip delay is about:
- 3 ms
- 6 ms
- 0.6 ms
- 60 ms
Show answer
Answer: B. One way = 600 ÷ 200,000 = 0.003 s = 3 ms. A round trip doubles this to 6 ms. The 3 ms figure is only the one-way delay.
-
5. The Mirai botnet of 2016 mainly exploited which weakness of IoT devices?
- Their use of quantum encryption
- Their high cost
- Their dependence on 1G networks
- Default or weak passwords that owners did not change
Show answer
Answer: D. Mirai scanned for cameras, routers and similar devices still using factory-default credentials, took control of them and used them in large DDoS attacks.
Sign in to keep track of the lessons you finish.