Author: Alan Gatherer, Founder and CTO, Cirrus360
The #oran (open RAN) Community has done a fantastic job disaggregating the RAN into components and automating the construction of RANs from these components. To do so we have drunk enthusiastically from the well of knowledge in the #cloudcomputing space. Now, as deployments are beginning to scale, we have become more interested in the performance and optimality of the components themselves. This is especially true for #private5g deployments where the use cases are highly specialized.
At Cirrus360 we believe the way forwards for component optimization is to again return to the cloud computing well of knowledge. Specifically we need to take note of the 2017 Turing Award given to Profs Hennessy and Patterson, which placed a mile marker in the history of computer science, recognizing that optimization for power and performance in the future would require the twin technologies of Domain Specific Architecture (DSA) and Domain Specific Languages (DSL). In 2017 the world of cloud computing and machine learning (ML) was already embracing this future, and this has created an explosion in the use of ML in many fields from biology to the stock market. The development of Software Defined Networks has also embraced the DSL/DSA philosophy, revolutionizing how we think about network deployment. Critically it has allowed for intent driven and automated network deployments.
Cirrus360 has developed a DSL for RAN component development and applied it to the development and optimization of the DU in ORAN. We have shown that an optimized DU can be specified in a hardware agnostic manner using an intuitive and intent driven DSL, and then the deployment of this code onto the DU can be automated. We believe this step can help RAN development achieve the same levels of system optimization and automation that we see today in SDN and ML development ecosystems.
Say hello to Alan on his Linkedin Post.