The open source community sees itself as an important connector between telecommunication operators that want to gain more cloud-like control over their expensive deployed assets and that cloud world those telecom operators want to emulate, which is a bridge nearing an important construction timeline.

Analysys Mason recently released a report that shed light on implementation challenges that could be preventing telecommunication operators from achieving a greater financial return on their immense 5G investments. It was based on a survey of 50 tier-1 operators, which found nearly all respondents believe open networks are critical to their survival though only two in 10 have an open network strategy in place.

“We are seeing that operators are really interested in building this more disaggregated and more open interfaces and more horizontal kind of network clouds for their 5G and potentially 6G networks. So interest and understanding is there,” Gorkem Yigit, research director at Analysys Mason, said in an interview. “But what we found out in this exercise was they are still struggling to realize these ambitions.”

One of the more pointed results of that report was the need for telecom operators to be more active in working with trade groups or organizations that are outside of the traditional telecom ecosystem.

“[Operators] do have high-level engagements, mainly on the traditional initiatives like 3GPP, TMForum, traditional telco kind of mindsets,” Yigit said. “But when we look at the ones that focus on more open networks, open clouds and things like that, the difference is quite huge.”

Linux Foundation works with operators for ‘open source networking’ The Linux Foundation is one of several groups that are looking to help surmount that challenge, taking advantage of its deep open-source community ties connected with its networking knowledge.

Arpit Joshipura, GM of networking, IoT, and edge, at the Linux Foundation, agreed with aspects of that report that noted telecom operators needed to be more active participants in non-traditional telecom groups in order to further the overall open network ecosystem.

“What we are saying is the next phase of this whole open journey is about not just everything that we have done, but it is about being able to collectively collaborate with open source,” Joshipura said. “What that means is, do I have the ability to co-innovate and speed up the deployment even more, because now I have access to code and I collaborate on all the software that is really non-differentiated, which happens to be 60% of the code on every part of the network.”

Joshipura said this means not caring if specific equipment is from vendors like “Ericsson, Cisco, Juniper. Who cares?”

Some of that progress can be seen in recent work around open APIs based on the Camara project that was launched by the Linux Foundation and in partnership with industry trade group GSMA. Joshipura explained that progress in this project is leading toward the next step, which he said was “open source networking.”

This path is being led by some forward-leaning operators that want to gain deployment speed and flexibility already enjoyed by the cloud ecosystem.

“They are defining how the right way to do open sources and they are contributing through their vendors, outside their vendors, they are focusing on things that they want fully open and partially open, and I think that gives us even more confidence that deployment happens,” Joshipura said.

And what about open telecom AI? Joshipura also pointed to a need for the development of more domain-specific artificial intelligence (AI) architectures. This means the need to scale current large language model (LLM) architectures to work in more diverse telecom environments.

“Every LLM or every learning or every AI implementation, whether it's enterprise or personal, is heading to a cloud, and it's going into a centralized data center. That is not scalable for edge applications, IoT,” Joshipura said. “That's not scalable for telecom domains where you do have an opportunity to get low latency, AI, [machine learning] learnings. That architecture and that work is probably the biggest focus we have right now.”

Joshipura explained that an example of this work is developing telecom data sets that can be used for training the network “without sacrificing privacy.” This includes a project that hides some data but still allows the model to learn.

“Then, boom, you utilize that and you get the domain-specific data sets and then you put it as a solution for everybody to learn, because this is not something that is going to be a competitive differentiator,” Joshipura said.

Joshipura also noted that the Linux Foundation was also working on trying to simplify network complexity across multicloud environments with its Paraglider project and trying to help cloud-native migration through its Cloud Native Telecom Initiative (CNTI).

“There are things that we can do to help accelerate deployments for vendors and solutions for the service providers,” Joshipura said.