Virtualising RAN And Core Networks At The Telecom Edge

Edge computing in telecommunications is changing where mobile network functions run, how quickly services respond, and how operators manage capacity. Instead of relying on a small number of central data centres, carriers are distributing compute, storage and network intelligence across radio sites, aggregation points and regional facilities.

This shift is closely linked to virtualised radio access networks (vRAN), open RAN, cloud-native packet cores and multi-access edge computing (MEC). Together, these technologies allow network functions to operate as software on standardised hardware, while placing selected workloads nearer to subscribers, machines and business applications.

For Australian operators, geography makes the question especially important. A service travelling between a regional mine, a mobile site and a Sydney or Melbourne data centre may face distance, congestion and backhaul costs that are less significant in denser European markets. Distributed architecture can improve responsiveness, resilience and local data handling when it is designed with operational discipline.

The Telco Stack Moves Beyond Central Data Centres

Traditional mobile networks were built around dedicated appliances and centralised core facilities. Baseband processing, mobility management, subscriber authentication and packet forwarding were tightly coupled to specialist hardware. This model provided predictable performance, yet it made capacity upgrades expensive and slowed the introduction of new services.

Virtualisation separates network functions from the hardware that hosts them. A virtualised network function may run in a virtual machine, while a cloud-native network function is generally decomposed into containers and microservices. The result is a more programmable telecom platform that can be automated through APIs, orchestration tools and continuous delivery pipelines.

In the RAN, compute-intensive baseband workloads can be divided between the radio unit, distributed unit and centralised unit. In the core, functions such as the access and mobility management function, session management function and user plane function can be placed according to latency, security and traffic requirements. This creates a continuum from central cloud to far-edge sites.

The architecture also supports network slicing, private 5G and service-specific traffic policies. A port operator might reserve predictable performance for autonomous vehicles, while a hospital campus could apply strict isolation to medical devices. These use cases depend on reliable orchestration across telecommunications, IT and operational technology environments.

Why Virtualised RAN Needs Edge Locations

RAN virtualisation is often presented as a route to lower hardware costs and greater vendor flexibility. Its practical value also comes from location. Fronthaul and midhaul connections between radio components have demanding timing, synchronisation and throughput requirements. Processing every signal in a distant cloud can create transport pressure and make performance harder to control.

A distributed unit positioned close to a cluster of cell sites can handle time-sensitive processing locally, while a centralised unit may serve a broader region. This split lets operators balance latency, utilisation and power consumption. It also supports coordinated radio features, provided the transport network has sufficient bandwidth and accurate timing.

Open RAN introduces open interfaces between radio components and encourages multi-vendor deployments. That can reduce dependence on a single supplier, though interoperability testing becomes a major engineering task. Operators must validate radio performance, synchronisation, accelerators, lifecycle tooling and fault management under real traffic conditions rather than relying on interface compliance alone.

In Australia, RAN placement has to account for long distances between population centres and difficult access to remote sites. A regional network serving Perth’s outer suburbs, inland communities or mining operations cannot assume the same fibre availability as central Sydney. Edge nodes can reduce the distance to users, but they also need rugged enclosures, remote monitoring, backup power and protection from heat, dust, storms and bushfires.

Cloud-Native Core Functions At The Edge

A virtualised 5G core gives operators greater control over where user-plane traffic is processed. The control plane can remain in a regional or national facility, while user-plane functions are deployed at metropolitan, campus or industrial edges. This arrangement reduces the path taken by high-volume application traffic and can keep sensitive workloads within a defined jurisdiction.

MEC extends this capability by hosting applications near the access network. Video analytics, augmented reality, industrial control, connected transport and gaming can use local compute without sending every packet through a distant core. Australians already expect rapid mobile access for streaming sport, contactless payments and cloud applications; edge platforms help operators meet those expectations when centralised routes become inefficient.

The industrial case requires particular care. A mine, factory or utility may still depend on programmable logic controllers and supervisory control systems that were never designed for cloud connectivity. Legacy control systems need segmentation, protocol-aware gateways and carefully managed update paths before an edge platform can safely exchange data with them.

This is where local private networks become valuable. A mining company in Western Australia could operate a private 5G network with an on-site user-plane function, allowing vehicle telemetry and machine vision to remain within the operation. The carrier may provide spectrum, orchestration and managed services, while the enterprise retains control over applications and operational policies.

Security, Resilience And Regulatory Boundaries

Virtualisation changes the security model of a telecom network. Dedicated appliances offered a relatively fixed boundary; software-defined infrastructure creates more interfaces, APIs, images, containers and orchestration components to protect. A compromise in the management plane could affect many network functions at once, making identity controls and privileged access management essential.

Operators should apply zero-trust principles between workloads, sites and administrative roles. Secure boot, signed images, hardware root of trust, vulnerability scanning and runtime monitoring help establish confidence in the platform. Network functions should be isolated according to their risk, with east-west traffic inspected rather than assuming that internal communication is safe.

Australian regulation adds practical obligations. The Security of Critical Infrastructure Act and its associated requirements are relevant to telecommunications entities and other critical sectors, while the Privacy Act governs the handling of personal information. Data minimisation, retention controls and clear accountability are important when subscriber, location and enterprise data are processed across distributed sites.

Resilience must also be engineered locally. Flooding can affect fibre routes and power infrastructure in Queensland and New South Wales, while bushfires can isolate regional facilities and damage access roads. A sound edge design uses redundant paths, local failover, tested backup power and the ability to operate in a degraded mode. Redundancy that has never been tested is an assumption, not a resilience strategy.

Comparing Deployment Choices And Business Trade-Offs

No single placement model suits every workload. A central cloud is efficient for large-scale analytics and common control functions, while a far-edge site is better suited to strict latency targets or disconnected operation. The correct decision depends on traffic patterns, service-level agreements, energy availability, compliance needs and the operator’s ability to manage thousands of distributed locations.

Energy is becoming a stronger consideration as well. Virtualisation can improve utilisation by sharing compute resources, yet additional servers, accelerators and cooling systems may increase local power demand. Facilities in hot Australian climates need efficient thermal design, and operators may need to coordinate deployment with renewable generation, battery storage and constrained grid connections.

Operators should also assess total cost over the network lifecycle rather than comparing server prices alone. Software licences, orchestration, site visits, energy, spares, cyber assurance and end-of-life migration can outweigh the initial hardware decision. A modular platform with repeatable site profiles is often easier to scale than a collection of bespoke edge installations.

Practical Priorities For Australian Operators

A successful programme starts with a clear service and workload inventory. Teams should identify which functions require ultra-low latency, which can tolerate centralised processing, and which must remain within a particular legal or operational boundary. This avoids moving workloads to the edge simply because the technology is available.

The following priorities can help shape a phased deployment:

• Begin with measurable use cases such as private 5G, local traffic breakout, industrial video analytics or network slicing.
• Standardise edge site profiles covering compute, storage, power, cooling, connectivity and physical security.
• Use automated provisioning, observability and remote lifecycle management for geographically dispersed locations.
• Test interoperability between RAN suppliers, accelerators, orchestration platforms and transport networks before production rollout.
• Map privacy, critical-infrastructure and data-residency obligations to each workload and operating partner.
• Build resilience around dual connectivity, local failover, backup power and procedures for prolonged disconnection.
• Track latency, packet loss, energy per workload, incident response time and total operating cost from the first pilot.

The skills model matters as much as the technology stack. Network engineers need familiarity with Kubernetes, APIs and infrastructure automation, while cloud specialists must understand radio timing, mobility and carrier-grade availability. Partnerships with universities, vendors and technical communities can help develop this combined expertise across Australia’s growing telecommunications market.

The Edge Computing Association can serve as a useful meeting point for operators, equipment providers, integrators, researchers and employers working through these changes. Join the conversation, share deployment lessons and connect with specialists building secure distributed networks across Australia and beyond.