Computer networks allow computers to communicate with one another, and provide the fundamental infrastructures supporting our modern society. Research on computer networks at Yale improves on essential network system properties such as efficiency, robustness, and programmability. The research spans all networking layers, including application-network integration (ANI); highly robust, flexible networking; software-defined networking (SDN) and programmable networking applications; and mobile networking.
Application-network integration (ANI): the traditional architecture of network-oblivious applications and application-oblivious networks is reaching its limit. Unilateral actions by applications alone or networks alone cannot efficiently support increasingly data intensive networked applications, with stricter requirements, in increasingly complex and heterogeneous network infrastructures that span edge and cloud. Yale Computer Science plays a leading role in the field of ANI. The research at Yale has led to the establishment of the Application-Layer Traffic Optimization (ALTO) Working Group of the Internet Engineering Task Force, and the definition of the ALTO Protocol, which is the first Internet standard supporting joint application and network interactions. Ongoing ANI work at Yale includes emerging fields such as ANI in 5G cellular networks, application-defined networking (ADN), and joint network-application programmable networking.
Highly robust, flexible networking: Although computer networks are becoming a critical infrastructure of our information-based society, they still have not reached a level of being highly reliable. The recent emergence of software-defined networking and learning based networking control has substantially improved network flexibility, but not network reliability. Yale Computer Science is leading the design, implementation and deployment of a novel networking control architecture based on multiple control plane composition, to achieve network reliability and programmability that cannot be achieved by individual, modular networking alone. Designed to address the challenges of tactical networks, the architecture can be fully realized in other modern networking infrastructures, with ongoing work realizing the architecture in some of the largest cloud data center networks and mobile networks.
Software-defined networking (SDN) and programmable networking applications: A major transformation of the modern networking infrastructures is the emergence of software-defined networking, which includes programmable networking data path, and the separation of the control plane and the data plane. The deployment of these techniques in some of the largest networks has been massively successful. Yale Computer Science plays leading roles in this important field, in designing high-level programming languages for SDN, joint network-application SDN programming, and novel applications of programmable data paths.
Mobile networks: Modern mobile networks are undergoing a radical transformation, migrating from dedicated, specialized infrastructures to public, shared ones, while delivering unprecedented throughput, coverage and latency. This transformation necessitates the virtualization of resources across the entire network stack, from spectrum to computation. It coincides with the emergence of edge data centers where latency-sensitive network functions must reside. Yale Computer Science has played a leading role in the development of massive MIMO technology and its virtualization for mobile networks and is a major participating university of the NSF AI Institute for Edge Computing Leveraging Next Generation Networks.