TL;DR:

  • 5G network slicing creates isolated, guaranteed-performance virtual networks on shared physical infrastructure — an industrial plant gets its own slice with reserved bandwidth, latency, and reliability, independent of what else is on the network
  • Multi-access Edge Computing (MEC) moves compute to the network edge (inside the factory, warehouse, or port), reducing round-trip latency from 50-100ms (cloud) to 1-5ms
  • Together, these enable applications that were impractical on Wi-Fi or 4G: machine control, automated guided vehicles with real-time coordination, and remote operation of heavy equipment

What Network Slicing Actually Does

A 5G network’s physical infrastructure — radio spectrum, base stations, core network components — is shared across many users and use cases. Without slicing, all traffic competes for the same resources. A video streaming spike could theoretically affect the latency experienced by an industrial sensor.

Network slicing uses virtualisation to create logically separate networks on the same physical infrastructure. Each slice has:

  • Guaranteed bandwidth: A minimum throughput that isn’t affected by other users
  • Latency targets: Maximum end-to-end delay enforced by traffic prioritisation
  • Reliability guarantees: Packet loss targets (5G’s URLLC — Ultra-Reliable Low-Latency Communication — aims for 99.9999% reliability at 1ms latency)
  • Isolation: Traffic in one slice is logically separated from other slices

For an industrial operator, a slice is functionally a private network that happens to share physical infrastructure with public mobile traffic. The guaranteed parameters mean they can make engineering commitments based on network behaviour, rather than treating connectivity as a best-effort variable.

The Three 5G Service Categories

The 3GPP standards body defined three service categories that align with different industrial use cases:

eMBB (enhanced Mobile Broadband): High throughput, tolerates moderate latency. Used for video surveillance, HD/4K camera feeds, augmented reality assisted maintenance. The familiar “fast 5G” that most consumers experience.

URLLC (Ultra-Reliable Low-Latency Communication): The category most relevant to industrial control. Sub-millisecond air interface latency, extremely high reliability. Designed for applications like machine control and autonomous vehicle coordination where missed packets have physical consequences.

mMTC (massive Machine-Type Communication): High device density, low data rates, long battery life. Designed for large-scale sensor networks — thousands of sensors across a facility, each sending small periodic packets.

Most industrial deployments use a mix: URLLC slices for control systems, eMBB for video and AR, and mMTC for environmental monitoring.

Multi-Access Edge Computing (MEC)

Network latency has two components: the radio access network (RAN) and the transport to wherever compute lives. 5G’s URLLC reduces RAN latency to sub-millisecond levels. But if application logic runs in a cloud data centre hundreds of kilometres away, the total round-trip might still be 50-100ms.

MEC places compute resources at the base station level or within the customer premises. A MEC server co-located with the 5G base station at a factory site means:

  • Local breakout: industrial IoT traffic never leaves the site
  • Round-trip latency of 1-5ms including compute time
  • Data sovereignty: sensor data stays on-premises, relevant for manufacturing IP and compliance

MEC servers run application logic and edge services: video analytics, machine vision inference, edge orchestration platforms, and data aggregation. The standard frameworks (ETSI MEC, 3GPP edge) define APIs for deploying applications to MEC nodes.

In practice, MEC hardware is often deployed by the mobile operator as part of a private 5G deployment, or procured by the enterprise for an on-premises 5G core installation.

Private 5G vs Public 5G Slicing

Industrial deployments use one of two models:

Public network slicing (operator-provisioned): The enterprise contracts with a mobile operator for a dedicated slice on the public 5G network. The operator provides SLA guarantees. This works for use cases where the site is covered by operator infrastructure and the reliability guarantees are sufficient.

Private 5G: The enterprise deploys its own 5G base stations and (optionally) its own core network on-premises. In the UK, Ofcom issued 3.8-4.2 GHz shared local access spectrum for this purpose. Private 5G gives complete control over the radio environment, eliminates shared spectrum contention, and allows on-premises data residency. It’s more expensive and requires radio engineering expertise.

Most large-scale industrial deployments — automotive plants, ports, warehouses above a certain scale — are private 5G. Public slicing is more common for mid-size deployments or where operator coverage already exists at the site.

Industrial Use Cases

Automated Guided Vehicles (AGVs) and autonomous mobile robots: AGVs in a warehouse or factory floor need to coordinate position and path planning in real time. Wi-Fi handoff between access points introduces latency spikes that cause vehicles to slow or stop; 5G with seamless mobility handles this without interruption. AGV fleets at port terminals and automotive manufacturers represent some of the earliest production 5G IoT deployments.

Machine control and closed-loop automation: Some manufacturing control loops require actuation within milliseconds of sensor feedback. Traditional approaches wired these connections to avoid latency variability. URLLC 5G allows wireless closed-loop control at latencies previously only achievable with Ethernet, enabling more flexible factory floor layouts.

Remote operation of heavy equipment: Mining operations, port crane systems, and construction machinery can be remotely operated from a control room with URLLC links. The latency and reliability requirements are such that consumer 4G/LTE is inadequate; URLLC at 1ms makes the feedback loop feel direct.

AR-assisted maintenance: Technicians wearing AR headsets receive overlay guidance from systems that can see their field of view via eMBB video streaming and respond with contextual information. Local MEC hosting the inference model keeps response times low enough for useful AR overlays.

Dense sensor networks for predictive maintenance: Vibration, temperature, and acoustic sensors on manufacturing equipment feed condition monitoring systems. 5G’s mMTC category handles the density (thousands of sensors) without spectrum congestion.

Practical Deployment Considerations

Spectrum: Private 5G in the UK uses Ofcom’s shared local access spectrum (3.8-4.2 GHz). Deployments require a licence application. The 26 GHz mmWave band offers higher capacity for dense indoor deployments but requires more base station density.

Integration with existing OT networks: Industrial facilities have existing operational technology (OT) infrastructure — PLCs, SCADA systems, industrial Ethernet. 5G doesn’t replace this; it adds wireless connectivity that integrates with it. The integration layer (translation between industrial protocols like PROFINET and IP, security boundary between OT and IT networks) requires engineering attention.

Security: 5G’s native security is stronger than Wi-Fi (mutual authentication, mandatory encryption on the air interface), but the expanded attack surface of connected OT equipment requires defence in depth beyond the radio layer. Network segmentation between the 5G core and OT networks, device certificate management, and monitoring for unusual traffic patterns are standard requirements.

Vendor landscape: Nokia, Ericsson, and Huawei provide end-to-end private 5G infrastructure. Smaller vendors (Celona, Druid Software, Casa Systems) offer more modular deployments. System integrators typically handle the OT integration layer. The market is fragmented and deployment expertise varies significantly.

For organisations evaluating 5G IoT, starting with a specific high-value use case — one AGV fleet, one production line, one crane — rather than a site-wide deployment is the typical path. This lets teams build 5G operational expertise before committing to infrastructure at scale.