What’s Next for Celona – Bigger, Better, Faster Next-Gen RAN
Mehmet shows how all these previous elements come together in our platform, and provides a glimpse into what to expect from Celona over the next year, including our next-generation RAN portfolio.
Presented by Mehmet Yavuz, CTO and Co-Founder. Recorded live at Mobility Field Day 13 in Santa Clara, CA on May 7, 2025. Watch the entire presentation at https://techfieldday.com/appearance/celona-presents-at-mobility-field-day-13/ or visit https://techfieldday.com/event/mfd13/ or https://Celona.io for more information.
Transcript
So again, in the, in the tradition of introducing something new and exciting, this is something that is fresh off the oven. We, uh, introduced, uh, it in beta to some of our recent customers. It's called Cel Orion, and it is our foray into AI based automation of the entire network lifecycle.
Alright, so just a quick thought of how we have actually been building this out. Uh, our data scientists are working very, very closely with our customer success, our engineering teams looking at all the issues that, most common issues that actually come up. And, you know, if you look at it, most of the common issues actually fall in these seven, eight buckets.
We looked at what our playbooks, our support team, our customer success teams, are actually using to solve these issues. All of that is actually being fed in as input to Cel Orion. And it is using that to automatically detect issues, do the analysis, do the root cause analysis, and actually present recommendations and remediation to customers.
And we obviously want to get to a point where we can actually present those solutions and present those remediations to our enterprise customers and in some cases actually take action to close the loop for our end customers. Now, I know self-healing and automation and all of that is a, is an abuse term and uh, it's been promised for a very, very long time. But I truly believe that we are at a point where this can be a reality, right?
And there are a few factors that are causing this. Number one, uh, you need to have enough data to be really able to make sense of and generate insights and generate actionable insights. And over the last six years we have been able to deploy across one 50 customers in different parts of the world, different verticals and uh, different sizes of networks, different shapes of networks, and that has given us the right to play in this particular space.
So that is number one. Uh, number two, I believe the AI tools that are available, and I'm sure all of you are using that today, the AI tools that are actually available to us today actually enable us to create very, very customized models that enterprises and partners who are actually managing these networks for our customers can actually use to automate a lot of the workflows that they do manually today, right? So I would say those are the two big factors that actually, and number three of course is given that this, there's a cellular network, there is a very deterministic way of how cellular network operates in terms of when does a device send data, when does it use the radio resources to send data?
When does the device actually hand over which access point it hands over to all of that it very, very closely managed by the network. And I think all of those controls actually allow us to deliver a truly automated autonomous network. And our goal is to look at, as I said earlier, all the aspects of the network lifecycle, whether it is network, bring up the RF design phase, the actual, uh, proactive monitoring of the network once it's up and running, being able to do the RCA and remediation.
And then once we get the confidence of the customers and we get their permission, be able to actually do autonomous resolutions of a lot of these issues that we actually see on the network in order to overall improve the end user and device experience. So again, I've rushed through a little bit, uh, so Cel Orion is available in the orchestrator. I'm gonna actually log in and give you a quick flavor of what that looks like and uh, and then we can open it up for any further questions.
So I'm logging into our, what we call as a beta network. This is our office network in Campbell Celonis, uh, office network in Campbell. And when a user now logs into the orchestrator, as I said, this is now available to all of our customers and we've been getting some really, really good feedback on, uh, on the capabilities that we are starting to show.
It's been laid out like a Kanban board, right? And our goal here is CEL Orion basically is going to instantiate agents and those agents are going to consume all of the data that is available on the network, whether it is device connecting, whether it's access point metrics, whether it is KPIs from the CEL Edge and core, and look at all of that and be able to analyze that, automatically detect issues, uh, tell the customers what action is needed in case there is some intervention required from them, and then move over to resolution state, either automatically or through a customer intervention. And our goal here is to create more and more of these tasks that are actually going to be fully automated using CEL Orion and reduce the number of human powered tasks that are actually being done.
So that's the goal. Again, we are starting down that path. It's not gonna be one big bang as you all know, it's gonna be a journey.
And here are some use cases of issues that are actually being detected on the network today, uh, autonomously. So here is a case where there was poor channel quality that was actually observed during a device communication. So this potentially, uh, points to some level of interference, whether it's external interference or internal interference.
There's a severity associated with it based on the number of devices that might have been impacted. It does a root cause analysis and provides potential recommendations in a as to what could be the source of issue. Is it inter is it truly interference or is it actually time synchronization that is causing these interference to happen and hence the performance degradation.
And you could look at all of this data from the perspective of the access point, from the perspective of the specific traffic direction, uplink, downlink or specific frequency, and see where or which frequencies are we seeing the worst performance in. Similarly, you can look at not just issues that are happening on the private 5G network. The system can also detect issues that could be happening on the enterprise network as, uh, MJ very nicely pointed out, you could actually get IP addresses for these devices from the salon network or you could get it using the enterprise DHCP.
And if for whatever reason the DHCP system within the enterprise is failing or the communication the DSCP server is failing, there's IP address scope exhaustion happening on the d CP network, all of those can actually be detected and actually put forth to our end customers to actually go and solve the issue rather than them having to go hunting for what the problem could be and what the resolution could be. There are others here. Here's an example of where the IP pool has actually been exhausted.
Uh, if the device has been misconfigured, if the device has been configured with an IP type that is not supported, the system can detect that the user can go and, uh, address those device configuration. When I say device, I mean a Zebra device or an Apple device, and actually go and solve that problem and move that to the resolution state. Uh, you could also have a scenario where a SIM is not activated.
That's a common problem that we see because this is something new for customers they don't know about SIM cards. So if a SIM has not been activated and device is trying to connect to the network, that can be flagged and that can be auto resolved if needed. And if there is a security issue in terms of a device actually connecting, which is not authorized to connect.
For example, if somebody were to take out a sim, put it on a different device, and now there is an IMEI mismatch, so this sim was actually tied to a specific device and now happens to show up on another device, all of those could be automatically flagged and then you could actually tie it to ServiceNow or any of your security sim systems to then trigger a specific workflow associated with that. So I'll pause here. There's a lot that we are working on here and is gonna continue to evolve and we'll Hopely have more to show in the upcoming months on Cel Orion.
But remember the name CEL Orion. Yeah. So with that, I'll probably hand it over to Mammoth to present the last section.
We have about six minutes and just want to give you a glimpse of what is coming, uh, in our innovation pipeline, right? So first of all, the edges architecture, our customers can't wait for it. There's a lot of demand.
So we'll be launching it, it will be commercial and uh, not only for private, but also for neutral host. This will really help us scale because it's not only cost, uh, of an appliance, but also the complex tea and so forth for those branch offices, small sites, even larger sites. Our, our customers really like the scalable model here and uh, that that's going to be, uh, an interesting, exciting journey for us for this, uh, rest of this year.
Uh, AI powered networks. I mean Orion is, is a key initiative for us. And one thing, uh, I don't know if you notice on the top left there was on the dashboard there was Orion powered and manual, uh, powered items.
So we constantly keep track of how many issues we resolve by with Orion and and that will be a big focus for us because this is not like a one time thing as as, uh, pre mentioned. It's an ongoing effort. And our goal is to constantly increase the number of issues, uh, that can be resolved with Orion, but that's only day end.
We are also expanding it to day minus one and day zero, right? How can we automate the whole RF design for a deployment? How can we, uh, automate to bring up and minimize the manual work in that?
So there's a lot of focus and effort going into that. 'cause we really want this to be truly the whole network lifecycle management, right? From the initial design to actual day and operation.
So we are thinking through the whole lifecycle and there will be a lot of exciting, uh, things that, that we'll hopefully share with you next time on that as well. Uh, we talked about neutral host and with one solution we can do both public and private as you have seen. And uh, this is the feature of indoor cellular in our mind, right?
We don't necessarily see this as two different things. And right now we are giving, uh, the option to the customers so they can start the journey with private and then add neutral host later. Or they can start with neutral host and then add private later.
But that, that is coming together nicely as one solution, right? So that, that will, that, that will continue as well. There are a few other things such as our high availability architecture HA architecture with the Kubernetes framework.
We start with the three node architecture, but we are moving to more standard two node HA architecture, which is more enterprise friendly I would say. So that will get, uh, commercialized very soon. Uh, on the zero trust it OT security, uh, our airlock I mentioned a bit, uh, this becomes more and more important as the OT use cases come, uh, in addition to it, how we segregate uh, both solutions, how we integrate with other firewall solutions out there and how can we bring additional enhanced, uh, solutions beyond honestly traditional wifi and other archival because we have key ingredients here such as SIM and uh, device hardware identity that we can really use to enhance the service.
So that's ongoing. We will keep expanding our, uh, footprint, uh, globally in different countries. To your point key, it's not only the private 5G spectrum countries, but also in any country really.
Now we have a model where we can go to the operators in that region and strike a deal with them and usually there's more than one interested to participate in this. 'cause honestly they see this as a growth. So, we'll, we'll keep expanding on that.
And last but not the list is our next gen ran. We didn't get a chance to talk about it, but we have a a, a further enhanced, uh, RAN coming in and we already see the benefits of our investment in ran because this, that was a big question, right? We started early on with the enterprise mindset, right?
We said, look, in the wifi world, you don't get APS from one vendor controller from another vendor. You want one vendor to give you all of those solutions. We started like that and we stuck to that.
And the benefits of edgeless that you have seen today, Keith, you were asking the right question, like what kind of processor is in there? When we designed the ap, we put the 4G modem, 5G modem and the processor, uh, in there because we knew we wanted to do the edgeless. We, we knew we wanted to put the data plane function in there so we put the right horsepower in the box.
That's why we are in, we are the only one who can do this today. If you just take an off the shelf AP like a 4G or 5G ap, you wouldn't have that horsepower to do the local breakout for the data traffic. So as we design the next gen, of course there is more MIMO, more spectrum carrier regulation, all of those things.
But there are advanced features that helps us with the timing, right? How can we minimize the, uh, dependency on external PTP time source and so forth. So we are thinking all of those things as we design the next gen ran.
I know if Tom is looking at me, we have one minute, so I'll maybe stop here and see if you folks have any questions. NextGen ran. Uh, my curious curiosity is definitely up.
Uh, can you share, uh, I guess, uh, I guess it's a cloud-based RAN implementation. Actually, no. Okay.
Okay. Our next gen ran, uh, is basically our, our mainstream solution. Uh, lemme put it that way, alright, is all integrated AP solutions still POE.
Okay. It has wider bandwidth, higher data rates. We can do four by four MIMO, all of those things with some other enhancements such as like, uh, in terms of timing and uh, so forth one, but that's our mainstream.
They're in parallel looking at more like or n type solutions because we see some use cases that may drive the need, but, uh, mainstream will continue plug and play QE powered AP type solution. Can you share, uh, which chip set vendor you're working with? I don't think we are ready to share that.
Okay. But I mean it is, uh, it is again, it, it fits to the principles I mentioned, right? And we've gone through a very detailed analysis of finding out the right chip set, uh, that meets all those requirements.
Alright, I can't wait On, on the installation. One of the things using CBRS, you need to have a certified person do it. You have to lock on the actual location in order for the SaaS to plain ice.
Yeah. Is there anything like internal GPS to help automate that process to make it easier for the installers? That's one of the enhancements, uh, we have in the works.
Right? Again, to automate that whole process. But I mean, one thing I have to mention, there is still an FCC requirement that somebody needs to sign.
Yeah. But if everything gets pre-populated and if somebody is just clicking, yes, this is location, this is location, this is location that's obviously going to help to streamline, right? Those are the kind of enhancements we are working on.