When designing next-gen RAN, there are three objectives that need be achieved simultaneously, including operational efficiency, automation of configuration and optimization activities, and customer level-specific service assurance. To achieve these 3 objectives, legacy RANs had RAN Resource Management (RRM); NetOps at the system level, and per user connectivity management. As next-gen RAN will have new elements including radios, people and things, all these devices will require configuration parameters and logs, which makes it difficult to perform these multitude of tasks via legacy RANs methods. In O-RAN, these tasks are divided into Near Real-time (10ms-1sec) and Non-Real-time (> 1sec) at the system level. O-RAN Alliance came up with RAN Intelligent Controller (RIC) to enable AI/ML based RRM optimization, and which is expected to be merged with the existing central Self-Organizing Network (SON) function. RT RIC is designed for controlling CUs and DUs through an open E2 interface. As CUs are placed between RT RIC and DU, the location of CU can be co-located with RIC or co-located with DU.
Near-RT RIC provides a software platform for xApps for RAN management and optimization. A large amount of network and subscriber data and Big Data, counters, RAN and network statistics, and failure information are available with L1/L2/L3 protocol stacks, which are collected and used for data features and models in Non-RT RIC. Non-RT RIC acts as a configuration layer to DU and CU software as well as via the E2 standard interface. They can be learned with AI and/or abstracted to enable intelligent management and control the RAN with Near-RT RIC. Some of the example models include, but are not limited to, spectrum utilization patterns, network traffic patterns, user mobility and handover patterns, service type patterns along with the expected quality of service (QoS) prediction patterns, and RAN parameters configuration to be reused, abstracted or executed in Near-RT RIC from the data collected by Near-RT RIC. In summary, Non-RT RIC feeds data collected from RAN elements into Near-RT RIC and provides element management and reporting. Near-RT RIC makes configuration and optimization decisions for multi-vendor RAN and uses AI to anticipate some of the necessary changes.
The O-RAN reference architecture is based on well-defined, standardized interfaces that are compatible with 3GPP to enable an open, interoperable RAN. RIC functionality delivers intelligence into the Open RAN network with near-RT RIC functionality providing real-time optimization for mobility and handover management, and non-RT RIC providing not only visibility into the network, but also AI-based feeds and recommendations to near-RT RIC, working together to deliver optimal network performance for optimal subscriber experience.
Recently, ATT and Nokia tested the RAN E2 interface and xApp management and control, collected live network data using the Measurement Campaign xApp, neighbor relation management using the Automated Neighbor Relation (ANR) xApp, and tested RAN control via the Admission Control xApp - all over the live commercial network.
Source: Nokia
AT&T and Nokia ran a series of xApps at the edge of AT&T’s live 5G mmWave network on an Akraino-based Open Cloud Platform. The xApps used in the trial were designed to improve spectrum efficiency, as well as offer geographical and use case-based customization and rapid feature onboarding in the RAN.
AT&T and Nokia are planning to officially release the RIC into open source, so that other companies and developers can help develop the RIC code.
Parallel Wireless is another vendor that has developed RIC, near-RT and non-RT. What makes their approach different is that the controller works not only for 5G, but also for legacy Gs: 2G, 3G, and 4G. Their xApps or microservices are virtualized functions of BSC for 2G, RNC for 3G, x2 gateway for 4G among others.
Source: Parallel Wireless
As a result of having 2G, 3G, 4G, and 5G related xApps, 5G-like features can be delivered today to 2G, 3G, and 4G networks utilizing this RIC including: 1. Ultra-low latency and high reliability for coverage or capacity use cases. 2. Ultra-high throughput for consumer applications such as real-time gaming. 3. Scaling from millions to billions of transactions, with voice and data handling that seamlessly scales up from gigabytes to petabytes in real-time, with consistent end-user experience for all types of traffic. The solution is a pre-standard near real-time RAN Intelligent Controller (RIC) and will adapt O-RAN open interfaces with the required enhancements and can be upgraded to them via a software upgrade. This will enable real-time radio resource management capabilities to be delivered as applications on the platform.
The RIC platform provides a set of functions via xApps and using pre-defined interfaces that allow for increased optimizations in Near-RT RIC through policy-driven, closed loop automation, which leads to faster and more flexible service deployments and programmability within the RAN. It also helps strengthen a multi-vendor open ecosystem of interoperable components for a disaggregated and truly open RAN.
In early August of 2020, Intel and VMware announced that that they will work with an ecosystem of telecom equipment manufacturers, original equipment manufacturers and RAN software vendors build programmable open interfaces that leverage Intel’s FlexRAN software reference architecture and a VMware RAN Intelligent Controller (RIC), to enable development of innovative radio network functions using AI/ML learning for real time resource management, traffic steering and dynamic slicing. Both companies believe that this will assist in optimized QoE for rollout of new 5G vertical use cases. In a statement associated with the Intel/VMware announcement, Vodafone Group’s Yago Tenorio, head of network architecture, said, “Seeing VMware and Intel work together to enhance the performance of general purpose computing platforms to run uncompromised RAN workflows, and to help simplify the complexities of product integration, has the potential to ensure Open RAN matures more quickly. It is also an enabler for faster developments in the [RIC] area. This move is central to the Vodafone initial Open RAN vision of partners that excel in our area, as well as benefit the wider community.”
At the end of August 2020, the Open Networking Foundation (ONF) has introduced a Software Defined Radio Access Network (SD-RAN) project, supported by AT&T, China Mobile, China Unicom, Deutsche Telekom and NTT, to develop, as an initial focus, an open source Near Real-Time RAN Intelligent Controller (nRT-RIC) that is compatible with the O-RAN architecture.
Named µONOS-RIC, it is a microservices-based SDN controller based on a version of ONOS (Open Network Operating System), the ONF’s SDN controller that is already being deployed globally: i.e. Comcast’s next-gen access network and SK Telecom’s 5G core.
The µONOS-RIC is supposed to conform to O-RAN Alliance architectures, with the E2 interface used between the µONOS-RIC and multivendor RAN elements (RU/DU/CU). This will eventually pull out all the SON and resource management from the RAN where it used to sit in the legacy RAN to xApps in the new, O-RAN based RIC architecture.
Google is involved in this project and plans to provide the cloud platform on which functionality such as the open NRT-RIC and xApps might potentially run. In addition, Intel and Facebook are providing the underlying technology to be used for the project’s developments to have it ready for trial in 2021.
And recently, ONF clarified the role of SD-RAN in building RIC as following:
1) ONF's SD-RAN will leverage the O-RAN architecture and vision for the nRT-RIC. SD-RAN will implement an open source nRT-RIC (called µONOS-RIC) and exemplar xApps including implementations for handover and load balancing, and will then prototype and work with operators to trial combinations of these open components with vendor proprietary (or other open versions) of RU/DU/CU, RIC and xApps.
2) SD-RAN will follow O-RAN specifications as they are developed and all extensions and learnings from building the system will be contributed back to O-RAN ALLIANCE, especially as it applies to Deliverable #1 - SD-RAN will prototype extensions to the data models for the E2 and A1 interfaces to enable flexibility and support for a broad set of xApp functions and applications, with a specific goal of enabling a robust spectrum of RRM and SON functionality within xApps.
In addition, the SD-RAN community will also be collaborating with the OSC, with the intent that interoperable nRT-RIC implementations will come out of both especially to complete:
1) Deliverable #2: The SD-RAN project plans to contribute an SDK for the northbound interfaces from the nRT-RIC towards the xApps to support and promote availability of interoperable xApps and rApps.
2) Deliverable #3: SD-RAN exemplar xApps will help demonstrate xApp interoperability between OSC NR-RIC and ONF µONOS-RIC.
In September 2020, Samsung and KDDI demonstrated network slicing involving a RIC to manage radio resources to guarantee required service levels. The companies managed network slicing on a network using Samsung’s virtualized core, RAN, and orchestration.
Source: Samsung
The TIP OpenRAN 5G NR Project Group recently launched the RAN Intelligence and Automation (RIA) subgroup to develop and deploy AI/ML based applications (as xApps) for a variety of RAN use cases including SON, RRM and Massive MIMO. TIP's RIA subgroup will use ONF's open source µONOS-RIC platform, leveraging open interfaces based on O-RAN specifications to assemble solutions for use case development, testing and deployment creating a deeper partnership between the TIP, ONF and O-RAN Alliance.
In conclusion, moving to Open RAN helps create a more innovative and cost-effective RAN environment, based on cloud economics. RIC is an essential component for delivering apps and services over a wireless network.
