Ibrahim Matta, Associate Professor of Computer Science, has been awarded a $560,000 National Science Foundation (NSF) grant to develop a new approach to Internet architecture. The grant will fund the development of a prototype for Recursive Internet Architecture (RINA), an innovative, clean-slate approach to computer networking that uses inter-process communications (IPC), a set of techniques for the exchange of data among multiple threads in processes running on one or more computers connected by a network.
According to Matta, this new approach to Internet architecture is cleaner than competing proposals. “Our main goal is to overcome the limitations of the current architecture, which has become overwhelmed by the growing scale of the Internet and changing requirements for security and manageability.”
RINA, based on the concept of recursion, makes manageability and security possible no matter how big the Internet grows. “Recursion is how to solve big problems by breaking them down into smaller, manageable components – a divide-and-conquer approach,” says Matta. “In our case, the recursive entity or building block is based on what is needed for inter-process communications, including security, to support the network applications that are running on top. With the current architecture, the entire Internet tries to do that; we are saying do it on a smaller scale using recursive building blocks and building from the ground up. Each block contains its own management and security controls, so if security is breached, it’s contained within that block. It doesn’t spread all over the place.”
Under RINA, application processes communicate via a distributed IPC facility that allows RINA to view each IPC facility as a private network. Matta’s project will design and develop the repeating structures of RINA, along with various policies seen as useful in support of security, multi-homing, mobility, and manageability. The developed prototype will be tested on small- and larger-scale test beds over the next four years.
“Today, everything on the Internet is visible to everyone, including addresses,” says Matta. “With recursive blocks, everything is contained; they are black boxes to people outside, who must enroll and authenticate to communicate inside these blocks.” RINA will be an opt-in architecture that is wholly compatible with the existing Internet. By design, users will be required to build and manage their own private network blocks.
“RINA might make obsolete efforts currently underway to patch the Internet, such as IPv6 (Internet Protocol version 6),” says Matta. “With RINA, we won’t need that many addresses because networks won’t have to accommodate so many users. You won’t necessarily want to communicate with everybody.”
Matta maintains that RINA is superior to competing designs, which are constrained by the limitations of the current Internet architecture. “The leaders in network design are going in this [RINA] direction,” says Matta. “Companies such as Google already are building their own networks from the ground up. Our proposal provides a complete architecture for doing this.”