In recent years, the number of cellular networks users has increased dramatically. At the end of last year, there were over five billion unique mobile subscribers around the globe. By the end of 2021, the volume of cellular data traffic produced by mobile devices rose ~ 50% (annual growth rate).
In addition to providing services to mobile devices, cellular networks support other diverse applications, such as real-time video, connected autonomous cars, distributed sensors, and smart manufacturing. The requirements for cellular networks to become more adaptive and capable are being pushed by the new use cases and applications.
To meet the new demanding requirements, a revolutionized cellular architecture was introduced aiming at giving the radio access network (RAN) a more prominent role, which was traditionally responsible mainly for data transmission from the user equipment (UE) to the core network for further processing. The new role of RAN will include more intelligence and autonomous decision models that can bring significant reduction in transmission times and grant higher QoS.4
Open Ran represents a revolutionary step towards RAN evolution. It stipulated a radical transformation and revolutionary approach of the RAN technology, from a design perspective, up to the operational model of the network, the Open Ran virtualized model is built on off-the shelf hardware and cloud software in multivendor environment with open interfaces.
Open RAN when deployed at the network edge will enhance 5G applications offerings in autonomous cars and IOT, Network slicing and bring us closer to secure over the air upgrades.
This promising RAN architecture that seems to get adopted worldwide is the one suggested by the O-RAN Alliance.It was initially launched by five major mobile carriers (Deutsche Telekom AG, Orange S.A., Telefónica S.A., TIM S.p.A. and Vodafone Group Plc) a couple of years ago, it is nowadays supported by over 160 companies (including 24 mobile operators across 4 continents) representing an extraordinary example of collaboration in area to defining together a new novel technical standard that will be applied and implemented between operators and suppliers around the world.
The O-RAN architecture should bring on table a new dawn of next gen services and applications to enhance offering of mobile operators, including but not limited to, virtual network slicing and dynamic spectrum sharing. O-RAN will revolutionize network development, deployment, and operation. It uses Cloud-RAN principles to leverage the software-defined implementation of wireless communications and networking functions.
O-RAN introduces new paradigms: an extensible RAN concept where third-party applications and services can be integrated into the platform in order to promote evolution of their capabilities in an agile fashion.
The O-RAN architecture is based on open specifications and disaggregation, breaking the RAN into multiple open and capable units, using cloud technologies to achieve scalability and reliability. The Open RAN movement applies to all generations of mobile technology and Open RAN applies to all future Gs.
This architecture embraces cloudification and network function virtualization techniques to perform the base-band function processing by disaggregated radio units (RUs), distributed units (DUs), and centralized units (CUs).
Mobile network operators (MNOs) could install their own RUs, but then lease on-demand computational resources for the processing of DU and CU functions from commonly available O-cloud server due to variations of the load over the day. Multiple MNOs share networking as well as computational resources in a multi-tenant network scenario.
Also, 5G and next generation 6G networks are introducing architectural transformations from an inflexible and monolithic system to a flexible, agile and disaggregated architecture to support service heterogeneity, coordination among multiple technologies and, rapid on-demand service deployments. The emerging open radio access network (O-RAN) framework provides virtualization, intelligence, and flexibility while defining open interface for network innovation.
Instead of legacy interfaces that are vendor-specific and controlled by key industry players, it defines open interfaces and an open architecture for innovation at all layers. It assumes that the data driven nature of cellular network management will lead to the creation of generic modules and interface for data collection, distribution, and processing. The non-real-time radio intelligent controller (non-RT RIC) and near-RT RIC are the most important functional components introduced by O-RAN. The former is hosted by the Service Management and Orchestration (SMO) framework.
As long as latency constraints are respected, the latter may be co-located with 3rd Generation Partnership Project gNB functions. The O-Cloud is an O-RAN compliant cloud platform that uses hardware acceleration add-ons when needed and a software stack that is decoupled from the hardware to deploy eNBs/gNBs as virtual network functions.
The SMO consolidates several orchestrations and management services, which may extend to pure RAN management such as 3GPP (NG-) core management or end-to-end network slice management.
In the sense of O-RAN, we could define main responsibilities of the SMO :
- Performance, fault, configuration, accounting, and security (FCAPS) interface to O-RAN network functions
- Large-timescale RAN optimization
- O-Cloud management and orchestration via the O2 interface, including resource discovery, scaling, FCAPS, software management, and create, read, update, and delete (CRUD) O-Cloud resources.
AI and Automation are the key ingredients
An open, modular, cloud-native network is not enough on its own to deliver the experiences enterprise customers need. To ensure low latency, high reliability and that massive numbers of machines can be connected and operated on the network, CSPs need to deploy AI and autonomous networks. Networks should be self-configuring, self-healing, and self-optimizing to enable zero-touch operations and zero-wait for new services. This will enable CSPs to give enterprises the best possible user experience. Enterprise services place higher requirements on network performance, and networks must be able to support highly differentiated connection requirements; real-time, online, one-stop, on-demand subscription and provisioning; and management of end-to-end SLAs.
If we talk about intelligence of O-RAN a key design consideration is the usage of AI, machine learning and automation frameworks.
The O-RAN based 5G and future 6G networks will incorporate artificial intelligence (AI) into the deployment, operation, and maintenance of the network for efficiently managing the network resources across the different applications and services. An example is the ability to manage multiple services with different quality of service on the same network where the decisions on how to allocate the resources are decided autonomously. The openness of the architecture, the introduction of new IT technologies as well as machine learning into the RAN offer great promise in terms of meeting the requirements.
Machine learning can be used to automatically and efficiently manage network resources in diverse use cases such as traffic steering, quality of experience prediction, and anomalies detection.
ML-based systems suffer from a special type of logical vulnerabilities that stem from the inherent limitations of the learning algorithms, which is why they are not free of vulnerabilities.
by Tomche Hristovski, Divison Manager, Telecom