MEC: What Multi-Access Edge Computing Actually Shortens
MEC moves the UPF and the application towards the RAN. It shortens transport, not the air interface — how 3GPP steers traffic to an edge host, what happens when the UE moves, and where deployments go wrong.
A latency budget has three parts: the air interface, the transport to wherever packets leave the mobile network, and the application itself. Multi-access edge computing attacks the second, by moving the UPF and the application host towards the RAN. It does nothing for the first — and that decides most MEC business cases.
What it shortens
Light in fibre covers about 200,000 km per second, so transport adds roughly 1 ms of round trip per 100 km before any switching or queueing. An application 800 km away costs about 8 ms that an edge host removes. An application already in a metro data centre 50 km away costs half a millisecond — and MEC buys almost nothing.
Measure the transport path before buying edge capacity.
The second benefit is less talked about and often larger: traffic that breaks out locally never crosses the backhaul and core, which matters for high-volume flows like video analytics.
How 3GPP steers traffic there
The standard does not run applications. It gets packets to them.
Local UPF and data network. The SMF selects a UPF using the DNN, the slice and the UE's location, so a session can anchor at an edge site and leave over N6 to a local data network.
Application function influence. An application provider can ask the core, through the NEF, to route specific traffic to a particular edge location — the mechanism behind "the app decides where it runs".
Discovery. Release 17 added DNS handling that resolves an application name to the nearest instance, and an application-layer architecture for edge enabler servers.
When the UE moves
An anchor near the RAN is only near while the UE stays put. Session and service continuity modes decide what happens next:
SSC mode 1 keeps the anchor. The session survives and the path gets longer.
SSC mode 2 releases the session and re-establishes it at a closer anchor — break before make.
SSC mode 3 establishes the new session before releasing the old one — make before break.
Modes 2 and 3 change the UE's IP address, so the application must handle relocation, including its own state. That is an application design question, not a network one.
Who specifies what
ETSI's MEC group — renamed from mobile to multi-access in 2017 — defines the edge host and its APIs; 3GPP defines how traffic is steered to it. A deployment needs both, usually from different vendors.
Where it goes wrong
Edge UPF, central application. Traffic breaks out locally and then travels to the central site anyway.
DNS resolving to the wrong instance. The anchor is local; the name resolves to a data centre 600 km away.
Relocation the application cannot survive. SSC mode 3 moves the session cleanly and the application loses its state.
The mental model
MEC shortens transport, not radio. About 1 ms per 100 km of fibre is the gain on offer.
3GPP steers; ETSI hosts. UPF selection and AF influence get traffic to the edge; the MEC platform runs what is there.
SSC mode decides mobility behaviour, and modes 2 and 3 push relocation onto the application.
Further reading
- 3GPP TS 23.501 — 5G system architecture, including edge computing support and SSC modes
- 3GPP TS 23.548 — 5G system enhancements for edge computing
- 3GPP TS 23.558 — architecture for enabling edge applications
- ETSI GS MEC 003 — MEC framework and reference architecture
- 5G core architecture, for the UPF and SMF behind local breakout
Practise this on 5G6GTech
Our companion site turns these topics into flashcards and quizzes.
- Drill 5G concepts with the flashcard deck (opens on 5G6GTech in a new tab)
700 standards-aligned 5G concepts on a spaced-repetition schedule, covering RAN, the core and service-based architecture, the physical layer and 5G-Advanced.
- Test yourself in the 5G practice quiz (opens on 5G6GTech in a new tab)
Timed multiple-choice questions across the same 5G domains, with explanations for each answer.
Related Articles
5G Core Architecture: The Service-Based Design Explained
How the 5G Core works: service-based architecture versus point-to-point reference points, the network functions and what each does, control/user plane separation, and how slicing falls out of the design.
AMF vs SMF vs UPF: Who Does What in the 5G Core
A side-by-side breakdown of the three central 5G Core functions: what each one owns, how they interact during registration and session setup, and why the split matters for slicing and edge.
5G Network Slicing: S-NSSAI, Slice Selection, and What Actually Gets Reserved
How 5G network slicing works end to end: S-NSSAI structure, NSSAI types, the registration-time selection procedure, and why requesting a slice is not the same as getting resources.
PDU Sessions in 5G: What They Are and How QoS Rides on Top
A PDU Session is the 5G Core's unit of data connectivity. Here's how one gets established, what lives inside it, and how QoS Flows map down to radio bearers.