Summary

Computing is a way of perceiving the world. With the development of AI, Metaverse and Blockchain, computing has become a de facto extension of human capability. The Internet architecture is no longer a good match to adapt to these new scenarios, so how do we design a new architecture that addresses today’s problems and will take us into the future with even greater success than the past? Our answer is conceptually simple: generalize the Internet architecture by replacing the focus on where – endpoint addresses of hosts – with what – integration of the computing power that users and applications care about. In this report, we have proposed a new Internet architecture called “Computing-Oriented Networking Architecture” (CONA) based on this simple change.

We have sketched an initial blueprint of this new CONA architecture, and identified a core set of research problems that must be investigated in order to develop and validate it. These intellectual challenges include scalability of effectively sensing computing power and network capacity, fast orchestration based on elastic scheduling, the efficiency of signatures for finely-granular signing and verification, usable trust models for data-centric security, network security and defense, content privacy and authentification, and fundamental communication theory that incorporates AI functionalities as an explicit part of the communication model.

In order to develop CONA from this initial blueprint into a proven and deployed architecture, we must grow a community to research and experiment with this new architecture at scale. We offer this report to the research community as a statement of the CONA architectural direction and research agenda, and invite others to join the effort to advance a computing-centric approach to some of the Internet’s most pressing problems in scalability, security, sustainability, and stewardship.