What Is Open RAN?

Open RAN, also written O-RAN or open radio access networks, is an approach to building the radio access network of a mobile system in which the network functions are split into separate components connected by publicly specified interfaces, so that equipment from different suppliers can interoperate. It contrasts with the traditional radio access network, where a single vendor supplied the radio head, the baseband unit, and the proprietary interface between them as one integrated product. The two ideas that define the approach are disaggregation, meaning the separation of previously bundled functions, and openness, meaning that the interfaces between those functions are specified in public documents rather than held privately.

The 3GPP standards already define a functional split of the base station into a central unit, a distributed unit, and a radio unit. Open RAN work builds on that split by specifying the remaining interfaces that 3GPP left open to implementation, most importantly the fronthaul link between the distributed unit and the radio unit, and by adding a control layer for network optimization. Most of that specification work is carried out by the O-RAN Alliance, an operator-led group formed in 2018, while national programs including a NIST effort on open radio access network measurement and testing address conformance and performance evaluation.

Disaggregation and Open Interfaces

In a disaggregated base station the radio unit handles the radio-frequency chain and the lower physical layer, the distributed unit runs the remaining physical layer along with media access control and radio link control, and the central unit terminates the higher layers and connects to the core network. The Open Fronthaul specification governs the link between the radio unit and the distributed unit, defining control, user, synchronization, and management planes over an Ethernet transport. Because the central and distributed units run as software on general-purpose servers, an operator can place them in a cell site cabinet, an edge data center, or a regional facility according to latency budget, an argument set out in early industry presentations such as an NTIA-hosted overview of open radio access networks.

The RAN Intelligent Controller

The control layer added by the O-RAN architecture is the RAN Intelligent Controller, which exists in two forms. The near-real-time controller operates on loops between ten milliseconds and one second and hosts applications called xApps that handle radio resource management tasks such as handover control, interference coordination, and slice-aware scheduling, communicating with the network functions over the E2 interface. The non-real-time controller sits inside the service management and orchestration framework, works on loops longer than one second, and hosts rApps that train models and set policy through the A1 interface. This structure is what makes machine learning a first-class part of the architecture rather than a vendor add-on, and research on efficient placement of a disaggregated RAN intelligent controller examines how controller instances should be distributed across radio nodes to meet those latency budgets.

Integration, Testing, and Security

Opening interfaces shifts effort from the vendor to the operator. Multi-vendor deployments require integration testing that a single-supplier network never needed, and open test and integration centers were created to certify conformance. The security position is mixed: published interfaces can be independently audited, but they also enlarge the attack surface, and the introduction of third-party xApps raises questions about isolation and privilege. A widely cited survey on O-RAN architecture, interfaces, algorithms, and security catalogs these considerations alongside the open research problems in control-loop design.

Applications

Open RAN has applications in a range of settings, including:

• Public 5G macro networks, where operators seek supplier diversity in the radio access network
• Private enterprise and campus networks built on commodity servers
• Rural and greenfield coverage projects with constrained equipment budgets
• Network slicing for industrial automation and other latency-sensitive services
• Neutral host deployments serving several operators from shared radio infrastructure
• Research testbeds for machine learning applied to radio resource management