Research Area 1 - Switch

UTC-based pipeline forwarding will be investigated as a solution to the following two major bottleneck problems:

• Using UTC (coordinated universal time, available from GPS and Galileo) with pipeline forwarding for solving the switch bottleneck problem – specifically, how to construct a multi-terabit (e.g., 40 terabit) switch in a single chassis.

• Using UTC with pipeline forwarding for solving the bandwidth mismatch problem between optical and wireless network, which is a link bottleneck problem. Proper solutions of the two bottleneck problems are critical for ensuring proper network operation and users’ satisfaction, and consequently, improving the IP networks services and business perspective.

The two bottleneck problems are solved using UTC-based pipeline forwarding, which is a method known to be optimal and independent of specific implementations. (Pipeline forwarding was invented by Henry Ford about 100 years ago for the assembly line in the car industry. Pipeline forwarding is still the most efficient manufacturing process today. All computers today operate using pipelines, a simple extension of Ford’s assembly line.) The motivation for pipeline operation is to provide efficiency, speed, and scalability. In this project, we will design a scalable time-driven IP switch whose switching operation is based on the pipeline forwarding of IP packets over global networks.

The proposed activities in this project will result in a detailed chassis design that shows how pipeline forwarding can solve the electronic switching bottleneck while optimizing physical space– specifically, packaging a 40 Tb/s (terabit per second) switch in a single chassis. Dynamically, this switch will have switching granularity that ranges from 1 Mbps (megabit per second) to a full channel capacity of 10 Gb/s (gigabit per second), and can be deployed from the core to the edges of the network.

Accordingly, the proposed activities will advance the discovery and understanding of the limitations of optical and electronic switching packaging. Beyond just packaging, though, the outcome of this project will have a broad impact on both IP packet-switched networks and on many services. Networks will be closer than ever to being outfitted with highly improved switching equipment that will satisfy the Internet’s insatiable need for capacity – doubling every 12 months or 1000-fold in 10 years.