What’s the difference between non-real time vs. near-real time RAN Intelligent Controllers?
The non-real time and near-real time RAN Intelligent Controllers are part of Open RAN architecture and they each have unique responsibilities in the network. One of the differentiators between the non-real time RAN Intelligent Controller and the near-real time RAN Intelligent Controller is the time frame in which it needs to take action based on the data it receives. Operators and vendors alike are looking at how that implementation will best happen. There’s general alignment of the role that the non-real time RAN Intelligent Controller will play, which carries over from previous technologies, like SON (Self-Organizing Network). It’s easy to see how that transfers into what a non-real time RAN Intelligent Controller would do.
There is more discussion in the industry around the near-real time RAN Intelligent Controller and whether it is needed. There’s also a discussion on whether there are other ways to implement it with a concept of a dApp, where applications run directly on the virtual RAN. There’s also some thought that you would need tighter integration with the CU/DU and the other elements of the RAN with the near-real time RAN Intelligent Controller because of the time domain that it’s working on. I think it will become an area of focus.
Initially, there will be a phased adoption of the non-real time RAN Intelligent Controller. We will see how it transitions and if it can execute similar functions to a SON. Then, we’ll start to see where else we can get performance enhancements and ask how we can optimize the network with a near-real time RAN Intelligent Controller. It’s going to be very interesting to see which of these technologies are implemented by operators in different ways. And there will be learnings that will be fed back into the industry. It may be several years before we see the direction coalesce from how the near-real time RAN Intelligent Controller can be helpful. But I do think you’ll see some implementation, validation, and lab testing over the next few years that’s going to give us a lot more insight.
What are the benefits of the non-real time RAN Intelligent Controller?
The non-real time RAN Intelligent Controller has a few different functions. One is to collect data from the network and store that data so that it can be used by the rApps or by Artificial Intelligence (AI), which is a key component of what the Service Management and Orchestration (SMO) architecture is bringing to RAN management. The non-real time RAN Intelligent Controller helps us to collect and store data so it can support the AI that trains the different models. Then, they can be implemented and utilized by some of the rApps. Another key function of the non-real time RAN Intelligent Controller is to provide that framework or platform that the rApps will run on. It’s the area where they have access to the network and to the data, which gives them the ability to interface the network and make configuration changes. The non-real time RAN Intelligent Controller works to both collect data and then implement policies into the network, where it takes action and configures the network. The interface to the rApps happens over the R1 interface, as it’s defined by the O-RAN specification. The non-real time RAN Intelligent Controller is a key element that enables the rApps to run. The rApps are ultimately what’s going to execute on a specific use case or provide intelligence into the network, but the non-real time RAN Intelligent Controller is making that possible through the different services that it offers to the rApps.
What are the benefits of near-real time RAN Intelligent Controllers?
The near-real time RAN Intelligent Controller has a lot of similar functions to the non-real time RAN Intelligent Controller, but it’s operating at a much lower latency time domain, so it needs to make decisions in less than one second or closer to a millisecond time frame. From a deployment perspective, you’ll probably see near-real time RAN Intelligent Controllers deployed closer to the edge of the network where it can support that type of time domain. And whereas the non-real time RAN Intelligent Controller is setting broad policies, the near-real time RAN Intelligent Controller is going to be making more short-term decisions that directly affect things like hand-over optimization or traffic engineering that interface closer to the elements of the Radio Access Network (RAN), whether it’s the CU/DU or directly to the radio unit itself. The near-real time RAN Intelligent Controller probably has a smaller reach given the latency that it needs to address, which is why it will tend to be deployed more at the edge.
What are the benefits of rApps?
The development of the rApps is a key component of the overall Open RAN concept, infrastructure, and ecosystem. The health of Open RAN and its adoption are dependent on a wide variety of rApps being developed and that needs to come from a broad set of vendors. It’s not reasonable for one vendor to develop a full scope of rApps that predict all of the different use cases that could be needed. We need a healthy ecosystem of app developers. Fujitsu has our own catalog of AI-powered applications that we’re developing and implementing as part of the open ecosystem so that each RAN Intelligent Controller will support multiple vendors. There may be an organic integration within our rApps that the same vendor is developing or other third parties. There will be a mix of some RAN Intelligent Controller vendors who will have their own rApps and then there will be independent rApp vendors who will also enter the market. This combination will give us many rApps and operators will reap the benefits with such a variety out there. But something for the RAN Intelligent Controller vendors to focus on is that they have to build this ecosystem. One of the key differentiators for any RAN Intelligent Controller is the ecosystem of rApp vendors that are interfaced, deployed, and validated with the RAN Intelligent Controller. There are some ways that the RAN Intelligent Controller vendors can enable that and encourage it. Having a strong interface, whether it’s through Application Program Interfaces (APIs) or Software Development Kits (SDKs), is going to be really important. The ease with which they enable the rApp vendors is going to encourage them to interface with that rApp. Once a RAN Intelligent Controller vendor has a broad ecosystem of rApps, it will allow them to generate more value for the operator. If you’re interested in learning more about RAN Intelligent Controllers with non-real time and near-real time options, as well as rApps, watch our video series.
