A deep-dive into the architecture, technology, and engineering principles that make fifth-generation wireless networks fundamentally different from all previous generations — from the radio air interface to the cloud-native core.
5G — the fifth generation of mobile network technology — represents the most significant transformation in wireless communications since the introduction of mobile data with 3G. Unlike incremental upgrades, 5G introduces entirely new radio technologies, a cloud-native core network architecture, and a set of capabilities that extend far beyond the smartphone use cases that defined previous generations.
Defined by the 3rd Generation Partnership Project (3GPP) under Release 15 and subsequent releases, 5G is designed to simultaneously serve three distinct usage scenarios: enhanced Mobile Broadband (eMBB) for high-speed consumer connectivity, Ultra-Reliable Low-Latency Communications (URLLC) for mission-critical applications, and massive Machine-Type Communications (mMTC) for large-scale IoT deployments.
5G is formally specified in 3GPP Release 15 (NSA), Release 16 (SA), and ongoing subsequent releases. The standard defines the New Radio (NR) air interface, the 5G Core (5GC), and all interfaces between network functions.
Enhanced Mobile Broadband delivers peak speeds up to 20 Gbps and sustained multi-Gbps throughput for smartphones, fixed wireless access, and high-definition video streaming.
Ultra-Reliable Low-Latency Communications achieves sub-1ms latency and 99.9999% reliability for applications such as autonomous vehicles, remote surgery, and industrial automation.
Massive Machine-Type Communications supports up to 1 million connected devices per square kilometre, enabling smart city infrastructure, agricultural sensors, and logistics tracking.
The 5G network is divided into two primary domains: the Radio Access Network (RAN) and the 5G Core Network (5GC). These two domains communicate over standardised interfaces — the NG interface connects the gNB to the 5GC's Access and Mobility Management Function (AMF) via NG-C (control plane) and to the User Plane Function (UPF) via NG-U (user plane).
5G Network Architecture — Service-Based Architecture (SBA) with cloud-native network functions
The 5G RAN is composed of next-generation Node B (gNB) base stations that communicate with user equipment (UE) over the New Radio (NR) air interface. The gNB consists of two logical components in the disaggregated RAN architecture: the Central Unit (CU) and the Distributed Unit (DU), with the DU further connected to a Radio Unit (RU) in the O-RAN (Open RAN) model.
The NR air interface uses Orthogonal Frequency Division Multiple Access (OFDMA) on the downlink and uplink, with flexible numerology (subcarrier spacing from 15 kHz to 240 kHz) defined by the parameter µ (mu). This scalable numerology enables 5G to operate efficiently across a wide range of spectrum bands and use cases, from low-latency URLLC to high-capacity eMBB.
| Numerology (µ) | Subcarrier Spacing | Slot Duration | Typical Use Case |
|---|---|---|---|
| 0 | 15 kHz | 1 ms | LTE-compatible, sub-1 GHz |
| 1 | 30 kHz | 0.5 ms | Sub-6 GHz eMBB |
| 2 | 60 kHz | 0.25 ms | Sub-6 GHz / mmWave |
| 3 | 120 kHz | 0.125 ms | mmWave URLLC |
| 4 | 240 kHz | 0.0625 ms | mmWave reference signals |
One of the defining technologies of 5G is Massive Multiple-Input Multiple-Output (Massive MIMO), which uses large antenna arrays — typically 64 to 256 or more antenna elements at the base station — to simultaneously serve multiple users in the same time-frequency resource through spatial multiplexing. This dramatically increases spectral efficiency compared to the 4×4 MIMO used in LTE.
Beamforming is the technique by which Massive MIMO arrays focus radio energy in specific spatial directions rather than broadcasting omnidirectionally. In 5G, both analog, digital, and hybrid beamforming architectures are employed. Digital beamforming offers the greatest flexibility, allowing independent beam control for each user, while analog beamforming is more practical at mmWave frequencies where hardware complexity and power consumption are constrained.
A 64-element antenna array can provide up to 18 dB of beamforming gain, effectively multiplying the cell's coverage range and enabling reliable mmWave links at distances that would otherwise be impractical due to propagation losses at high frequencies.
5G is deployed across two fundamental spectrum categories, each with distinct propagation characteristics, capacity, and coverage trade-offs. Understanding the difference between these bands is essential to understanding 5G's capabilities and limitations.
Frequency Range 1 covers bands below 6 GHz, including low-band (600–900 MHz) and mid-band (2.5–4.9 GHz) spectrum. Sub-6 GHz 5G provides broad coverage comparable to 4G LTE, with peak speeds of 1–3 Gbps. The mid-band "sweet spot" around 3.5 GHz (n78 band) is the dominant global 5G deployment band.
Frequency Range 2 covers millimetre-wave spectrum from 24 GHz to 100 GHz, with current deployments primarily in the 24–29 GHz and 37–40 GHz ranges. mmWave delivers extreme speeds but has very limited propagation range and is severely attenuated by walls, foliage, and even rain.
The 5G Core represents the most significant architectural departure from previous generations. Rather than a traditional hardware-based network with dedicated nodes, the 5GC is built on a Service-Based Architecture (SBA) where all network functions are software-defined microservices that communicate over HTTP/2 using RESTful APIs. This cloud-native approach enables deployment on commodity hardware, rapid scaling, and continuous software updates without service interruption.
| Network Function | Abbreviation | Role |
|---|---|---|
| Access and Mobility Management Function | AMF | Handles UE registration, connection, and mobility management |
| Session Management Function | SMF | Manages PDU sessions, IP address allocation, and data path setup |
| User Plane Function | UPF | Forwards user data packets between RAN and data networks |
| Policy Control Function | PCF | Enforces QoS, charging, and traffic policies |
| Unified Data Management | UDM | Stores subscriber profiles, authentication credentials, and plan data |
| Authentication Server Function | AUSF | Performs subscriber authentication using 5G-AKA or EAP-AKA' |
| Network Repository Function | NRF | Service discovery and registration for all network functions |
| Network Slice Selection Function | NSSF | Selects the appropriate network slice for each connection |
Network slicing is one of 5G's most transformative capabilities. It allows a single physical network infrastructure to be partitioned into multiple independent logical networks (slices), each with its own dedicated resources, topology, and service characteristics. A slice for autonomous vehicle communications can be configured with ultra-low latency, while a slice for IoT sensors can be optimised for low power and high device density.
Each slice is identified by a Single-Network Slice Selection Assistance Information (S-NSSAI) value, which includes a Slice/Service Type (SST) and an optional Slice Differentiator (SD). The NSSF selects the appropriate slice at connection time based on the UE's subscription and request.
In NSA mode (3GPP Option 3), 5G NR provides additional radio capacity while the 4G LTE Evolved Packet Core (EPC) handles all control-plane signalling. This allows operators to deploy 5G quickly by reusing existing 4G infrastructure. The UE connects to both an LTE base station (eNB, acting as Master Node) and a 5G gNB (Secondary Node) simultaneously via Dual Connectivity (EN-DC).
In SA mode (3GPP Option 2), both the 5G NR radio and the 5G Core Network are deployed. This is the target architecture that enables the full capabilities of 5G including network slicing, ultra-low latency, and the complete SBA. SA deployment requires greater upfront investment but delivers the full 5G value proposition and supports advanced use cases that NSA cannot.
Multi-access Edge Computing (MEC), standardised by ETSI, extends the 5G architecture by bringing compute and storage resources to the edge of the network — physically co-located with or close to the base station. This architecture dramatically reduces the round-trip latency for applications that would otherwise need to traverse the core network and reach a centralised cloud data centre.
In the 5G architecture, MEC is realised through the deployment of edge UPF nodes and local Data Networks (DNs). The SMF can configure a PDU session to be "broken out" locally at an edge UPF, ensuring that traffic destined for a local edge application never leaves the local network domain. This is critical for applications such as real-time video analytics, cloud gaming, and industrial control systems.
The combination of 5G's sub-1ms air interface latency and MEC's localised processing can achieve end-to-end application latencies below 5ms — a threshold required for haptic feedback, real-time control systems, and extended reality (XR) applications.
5G introduces significant improvements to network security compared to its predecessors. The Subscription Permanent Identifier (SUPI), which replaces the IMSI used in 4G, is never transmitted in clear text over the air. Instead, it is encrypted using the Home Network Public Key (HNPK) to produce a Subscription Concealed Identifier (SUCI), preventing IMSI catchers (fake base station attacks) that were a vulnerability in 2G, 3G, and 4G networks.
Authentication is performed using 5G-AKA (Authentication and Key Agreement) or EAP-AKA', with the AUSF and UDM handling credential verification. The security architecture also introduces enhanced protection for the Non-Access Stratum (NAS) and Radio Resource Control (RRC) signalling, with mandatory integrity protection for all user plane traffic in certain scenarios.
Understand how devices gain internet access through mobile networks, from PDU session setup to IP routing.
Learn what "recharge" means technically — how quota is allocated and enforced in mobile networks.
Explore how 5G coverage is planned, measured, and experienced by end users in real-world conditions.