The evolution towards O-RAN (Open-RAN) architecture stands as a transformative milestone in the telecommunications industry, promising increased flexibility, cost-efficiency, and innovation. In this article, we will dive into the O-RAN architecture to familiarize ourselves with the components and relationships that enable a virtualized, intelligent, open, and interoperable RAN (Radio Access Network).
Source: O-RAN ALLIANCE e.V
Introduction
In our previous article, we introduced O-RAN, a telecom industry initiative that aims to revolutionize the current RAN (Radio Access Network) architecture by embracing intelligent software-defined architectures and introducing openness.
As we highlighted in the previous article, O-RAN has the potential to improve network reliability, performance, management, and adaptability and contribute to a more diversified ecosystem that leads to specialized solutions/providers and lower TCO (Total Cost of Ownership) for operators.
To better understand the O-RAN promise, we will introduce the O-RAN architecture, highlighting its components and interfaces that enable the Radio Access Network (RAN) to be virtualized, intelligent, open, and interoperable.
The O-RAN Macro Architecture
The following image highlights the macro architecture of O-RAN. Over the next sessions, we will introduce each component and the interfaces that enable their interaction. We will go through the architecture in a bottom-up approach, starting with the infrastructure layer (O-Cloud).
Note that although O-RAN leverages interfaces already defined by 3GPP such as E1, F1 and Open Fronthaul (used for communications between the network components), we will focus on the O-RAN interfaces.
O-Cloud
O-Cloud is the cloud computing platform of the O-RAN architecture. It is composed by:
- Pool of physical infrastructure nodes, that meet the O-RAN requirements to host virtualized network functions. The nodes can be distributed in one or multiple locations running COTS hardware (Commercial-Off-The-Shelf).
- Supporting software components such as the Operating System, Virtual Machine Hypervisors, Container Runtime, etc.
- Related management and orchestration components that enable the provisioning and orchestration of the infrastructure and virtualized network functions.
In summary, the O-cloud is the foundational layer for the O-RAN architecture; focusing on its virtualization paradigm, it enables hardware and software decoupling and resource sharing and automates management/infrastructure tasks.
O-RAN Network Functions (O-RU, O-CU, and O-DU)
O-RAN adopted the functional split 7.2x proposed by 3GPP. It features the main components Radio Unit (RU), Distributed Unit (DU), and Centralized Unit (CU) and was designed to strike a balance between the simplicity of the RU and the data rates and latency requirements on the interface between the RU and DU. In the O-RAN architecture, these components were renamed to O-RU, O-DU, and O-CU to include the “Open” prefix.
The image below and the subsequent text provide some details on the responsibilities of each component within the functional split:
Source: Based on image from Understanding O-RAN: Architecture, Interfaces, Algorithms, Security, and Research Challenges
The O-RU is typically deployed at the cell site and is responsible for interfacing with end-user devices. It processes Radio Frequency (RF) signals and handles the lower part of the Physical Layer (Low-PHY).
The O-DU, usually located closer to the cell site to meet latency and environmental requirements, manages communication with multiple O-RUs. It oversees the upper part of the Physical Layer (High-PHY), Medium Access Control (MAC), and Radio Link Control (RLC).
The O-CU, which can be positioned further within the network, facilitates higher-level functions, and serves as a gateway between the radio access network and the core network. It manages protocols such as Packet Data Convergence Protocol (PDCP), Service Data Adaptation Protocol (SDAP), and Radio Resource Control (RRC) entities.
Within the O-CU, an additional split occurs in two functional entities: the Control Plane (O-CU-CP) and the User Plane (O-CU-UP), as defined in 3GPP TR 38.806. This split enables independent scaling of control and user plane functions, enhancing flexibility and efficiency in network management and resource allocation.
In the O-RAN architecture, the O-CU-CP, O-CU-UP and O-DU represent virtual network functions deployed on top of the O-Cloud. The O-RU, however, is not yet virtualized according to the latest O-RAN architecture description (O-RAN.WG1.OAD-R003-v11.00).
Service Management and Orchestration (SMO)
The SMO consolidates the management services related to the Radio Access Network:
- FCAPS (Fault, Configuration, Accounting, Performance and Security) of the O-RAN network functions.
- RAN control and optimization.
- O-Cloud management, orchestration, and workflow management.
The SMO also hosts the Non-Real-Time RAN Intelligent Controller (Non-RT RIC) and is responsible for managing the array of applications operating within the RAN Intelligent Controllers, a topic we will explore in the following section.
In summary, the SMO serves as a centralized hub for managing all aspects of the RAN domain.
Ran Intelligent Controllers (Non-RT RIC and Near-RT RIC)
One of the crucial components of the O-RAN Architecture is the Ran Intelligent Controllers (RIC). As the name implies, it aims to bring intelligence to the RAN operation. These controllers can collect telemetry data from RAN components and execute control actions based on data analysis. By leveraging AI, they enable RAN control and optimization through closed feedback loops. To add to that, these controllers are extensible, meaning they can work as software platforms to hold multiple custom logic applications that handle different tasks or scenarios.
There are two of those components in the O-RAN architecture:
Near-Real-Time RIC (Near-RT RIC):
- The Near-RT RIC is deployed as a virtual function on the O-Cloud platform and performs management and control of the network at near-real-time, in feedback loops of 10 ms to 1 s.
- It holds xAPPs, which are custom logic applications responsible for collecting information, computing it, and outputting control actions to the O-RAN network functions for different applications such as traffic steering, QoS based resource allocation, beam mobility management, and others.
- Its control over the O-RAN network functions is steered by policies, ML models, and enrichment data provided by the other controller, the non-RT RIC, detailed below.
Non-Real-Time RIC (Non-RT RIC):
- As mentioned earlier, the Non-RT RIC is a component of the SMO framework.
- It supports intelligent RAN optimization by providing policy-based guidance, ML (Machine Learning) model management and enrichment information to the Near-RT RIC.
- It can also support RAN optimization in a non-real-time interval (control loops > 1s).
- It hosts custom logic applications called rAPPs that provide value added services to support and facilitate RAN optimization and management tasks.
The use of extensible RIC capabilities through custom applications (xApps and rApps) is one of the most interesting concepts of the O-RAN architecture. This facilitates the inclusion of tailored applications developed for specific use cases and scenarios, enabling flexibility, facilitating innovation/experimentation and also opening the market to new specialized players.
XApps and rApps can be used to quickly integrate innovation in ML and AI, continuously improve network reliability and performance, and provide agility.
O-RAN Interfaces
Another central concept of the O-RAN architecture is the Open Interfaces, enabling seamless integration among components from diverse vendors. It empowers operators to tailor solutions precisely to their requirements. As mentioned earlier; by adopting this approach, the market becomes more diversified and specialized, freeing players from the necessity of delivering complete stack solutions for the Radio Access Network. Instead, they can concentrate on developing specific components.
The O-RAN specifies the following interfaces to enable communication between its major components:
Interface |
Between |
Used for |
O1 |
SMO O-RAN network functions Near-RT RIC |
Enables the SMO to perform FCAPS of O-RAN network functions (excluding O-RU) and the Near-RT-RIC
* Fault, Configuration, Accounting, Performance and Security |
O2 |
SMO O-Cloud |
Enables the SMO to manage and orchestrate the RAN infrastructure running on O-Cloud |
Open Fronthaul M-Plane |
SMO O-RU |
Enables the SMO to perform FCAPS functionality of O-RU
* Fault, Configuration, Accounting, Performance and Security |
E2 |
Near-RT RIC |
Enables the Near-RT RIC to collect data from the O-RAN network functions and provide actions |
A1 |
Non-RT RIC Near-RT RIC |
Used by the Non-RT RIC to provide policies, enrichment information and ML models to the Near-RT RIC and collect feedback. |
Conclusion
In essence, O-RAN represents more than just a technological advancement; it signifies a paradigm shift towards a more agile, adaptable, and inclusive telecommunications landscape that can lead to an improved, more efficient, and cost-effective Radio Access Network.
In this article, we provided a quick overview of the O-RAN architecture components and interfaces to help you familiarize yourself with O-RAN terminology and concepts.
More details about the architecture can be obtained directly from the specifications provided by the O-RAN Alliance.
Acknowledgement
This piece was written by Guilherme Carrenho, Innovation Expert at Encora. Thanks to Douglas Lopes Pereira, Guilherme Batista Leite and João Caleffi for reviews and insights.