Ericsson has launched of a new software platform designed to enable operators to manage and optimise multi-vendor 4G and 5G radio networks, and to host third party software from applications developers.
The vendor’s Intelligent Automation Platform includes a Non Real Time Radio Intelligent Controller (Non RT-RIC), supporting O-RAN defined interfaces to radio baseband units, and to a near real time RIC (nrtRIC). It also comes with an SDK produced by Ericsson that third party developers and operators can use to integrate their own applications onto the RIC. (Applications on the Non-RT RIC are known as rApps, their counterpart on the near real time RIC are called xApps.)
Santiago Rodriguez, Head of Strategic Projects, Ericsson, said that the new platform is “based on Open RAN principles for Service Management and Orchestration (SMO)”.
Multi-vendor support
He said that the platform would be able to manage both Ericsson RANs and those of other vendors, including integrated (Ericsson calls it “purpose built) equipment that is already deployed in the field.
“We will provide interfaces to 4G and 5G networks from multiple vendors. So it should be possible to interface with the vendors’ Element Management System and allow the platform to automate those technologies as well.”
That would, of course, require the vendor to open up their management interfaces to the Ericsson RIC, something that the operator would no doubt have to prevail on them to do. A recent blog post from Ericsson’s Justin Paul seemed to be more cautious about how this will work. It concluded that “more work has to be done to see whether that concept [SMO] can be extended from Open RAN to the existing purpose-built RAN that makes up over 98 percent of deployed networks today.”
However, another post published two days later from Paul and Rodriguez takes an extended view and proposes a solution. It asked, “What would happen if you could take the SMO concept and extend the automation capabilities brought about by SMO … into the domain of the existing, purpose-built RAN networks? At Ericsson, we believe this is very much possible. To achieve this, we will start with the existing C-SON and network design and optimisation (NDO) applications already deployed in networks. Today, these applications are often deployed on tightly integrated application platforms to perform specific automation operations. The SMO architecture effectively allows you to recreate these applications as RAN automation applications or rApps, as well as standardising the underlying application environment: the SMO’s non-RT-RIC, which uses an R1 interface that enables interoperability between platforms and vendors.
“The ability to run rApps to optimize both Open RAN and existing purpose-built 4G and 5G networks is profound. If this can be achieved one of the major barriers to adoption of the SMO will be crossed – scalability.
“Obviously, it’s hard to justify investment in an automation platform for a tiny percentage of your current network. This is why we are expanding the scope of the SMO to cover Open RAN and purpose-built RAN. – this results in the investment, covering the entire network with the benefits of automation being similarly applied to the entire network.”
The article proposes that vendors open proprietary interfaces from their existing OSS – say Ericsson Network Manager or Nokia’s NetAct – into the SMO. Integrating this alongside control of the O-RAN elements that sit on open interfaces, would create a “single pane of glass” for automation of RAN optimisation.
Adding rApps to the platform
Rodriguez said to TMN in an interview, “The problem the IAP tries to address is that in networks today, service providers have multiple vendors and technologies, and with Open RAN comes more disaggregation and more complexity. So the platform supports multi-vendor access, multi-technologies in 4G and 5G, and virtual and physical networks. It supports open interfaces like O-RAN’s O1, O2 and A1, and it is open for innovation to run rApps that are not only from Ericsson, and they’re not only working on Ericsson radio. They can be from a third party software vendor, or a customer themselves, developing their own applications, whatever they choose to have.”
The primary goal of the Non RT-RIC is to support non-real-time radio resource management, higher layer procedure optimisation, policy optimisation, and provide parameters and AI/ML models to support the operation of nrtRIC functions in the RAN. Functions include service and policy management, RAN analytics and model-training for the near-RealTime RICs.
Ericsson has defined four application areas for the Non-RT RIC – deployment, healing, evolution and optimisation – and has initially ported three of its own network intelligence applications over to the platform, Rodriguez said. The rApps are for Frequency Layer Management, Centralised 5G Automatic Neighbour Relations and Performance Diagnostics. The initial rApps demonstrate a “proof-of-concept” that rApp development can be accelerated using the Ericsson SDK. Each rApp can interwork with other rApps to build more complex automation functions.
More would follow from Ericsson, but Rodriguez wanted to stress that the aim is also to encourage, via the SDK, third party innovation on the RIC.
That said, the details of how software vendors can engage to integrate their rApps are still to be worked out, Rodriguez said.
“So we’re going to start by offering an SDK , including template examples and an API interface description, that we are offering to partners as well as to rAPP developers. The ambition is that we will open up to more and more vendors. We are not decided exactly on all the details of the got-to-market, that will come later as we mature.”
With Ericsson producing its own SDK for its IAP, does that mean that software developers face having to integrate their software a number of times to different vendor RICs?
“There will be a level of standardisation via the R1 interface [see diagram above] and we will try to make that as simple as possible. At end of day it will be a little bit like developing an app on IoS or Android, there’s always differences in how you use the platform but the more we all rely on the open standards the better.”
Focus on higher layer
Since the beginning of its involvement with O-RAN, Ericsson has made clear it sees the most value in focussing on the upper layer of automated management rather than get involved, as Nokia has done, in the nrt-RIC specifications. This blog post from a year ago outlines why that is – focussing on the wider availability of contextual information at the higher layer.
Rodriguez said, “We think the SMO is a very attractive area for our customers and end users, the place where can drag bigger innovation. We think on this IAP where we have one second to react to what’s going on in the network, plus access other information outside of the network, we think this is the area where there can be more innovation, where we can have a much more open ecosystem and innovation. The lower you go the more integrated you get into the actual RAN.”
He added that the wider context of the launch is that customers are asking for solutions that can help them improve the performance of their networks in an intelligent and dynamic manner. One operator providing quotes to support the Ericsson press release is BT, which last week held an event on its Open RAN work, where once exec described the RIC as offering the most exciting opportunities within Open RAN. BT said it is developing its own rApps, as well as hoping to foster innovation from other vendors.
Asked if the operator thought that vendors would open up their radio layers to automated management from the RIC, BT’s Chief Architect Neil McRae, said, “Watch this space.”
Well, it didn’t take long for the space to be partially filled – and we’re watching. With this launch Ericsson has reiterated that it is open to being open at a higher level of management and orchestration. However, one area where there is still open space is its own support for interfaces such as E2, between RAN nodes and very time sensitive radio resource management apps that sit on the nrtRIC, where its rival Nokia has committed to developing support.
One question that may determine development in that direction will be how much value operators see in being able to access third party software on the nrtRIC, for example in areas such as mMIMO beam management and scheduling, versus the tightly coupled solutions that vendors traditionally provide. For now, Ericsson is banking that the high layer, C-SON type environment, is ripe for innovation and can bring operators the most potential for performance improvements.