Columbia University & GENI: Embedding Real-time Substrate Measurements for Cross-layer Communications (ERM)
The GENI ERM team at Columbia University, led by Prof. Keren Bergman, Director of the Lightwave Research Laboratory, is creating innovative cross-layer measurement capabilities that will help free GENI researchers from the constraints of today’s strictly layered Internet architecture.
By providing deeper exposure to physical layer measurement and protocol information, the ERM team will help enable GENI researchers to realize well instrumented experiments supporting new network architectures that cut across today’s protocol layers. The ERM team has developed initial specifications for cross-layer communications that will allow bi-directional information exchange between protocol layers and enable holistic optimization of the network stack from optical network substrates all the way up to the application, and are now working with other GENI prototyping projects to integrate their design into the overall GENI control and measurement framework.
To enable this top-to-bottom communication, the ERM team is developing GENI-compatible, real-time embedded measurement capabilities; enabling deeper exposure and transparency to network substrates for cross-layer information exchange and user access; and integrating their prototypes with GENI’s ORCA control plane to dynamically monitor optical substrate performance and allow for cross-layer control and management decisions based on the optical layer configuration and performance.
Prof. Bergman’s team has already achieved several important milestones. The team first evaluated the GENI system design’s capabilities to embed real-time measurements and assessed requirements for real-time user access to data measurements across a diverse set of heterogeneous technologies. Then they developed two sets of specifications; one for supporting real-time measurements within the network substrate and another for network protocols that access cross-layer measurements. They have also developed models using the ns-2 simulator to quantify the performance improvements that their techniques enable. Finally, the team has proposed a Unified Measurement Framework (UMF) that includes well-defined interfaces between the GENI control plane and measurement devices.
Now the ERM team is collaborating with the GENI prototype projects at the Renaissance Computing Institute (RENCI), Duke University, and the University of Houston for potential integration of their real-time measurement system into the ORCA-BEN control framework. Before the end of the year, the ERM team expects to contribute an experimental use case to GENI working groups to further validate the proposed cross-layer communications schemes.