STI: Development of Shared-Network Protocols and Strategies for e-VLBI

Massachusetts Institute Of Technology, Cambridge MA

Investigators

Abstract

Very-Long-Baseline Interferometry (VLBI) is one of the most powerful techniques available for the high-resolution imaging of distant radio sources in the universe and for making accurate measurements of the motion of the earth in space. Multiple radiotelescopes scattered over the surface of the earth simultaneously record data from a radio source at streaming data rates as high as 1 Gbps for a 24-hour period; the data are then shipped to a central processing site for correlation analysis. Because the signal-to-noise ratio achieved by VLBI increases with the bandwidth of the observations, there has been a continual effort to increase the captured data rate. The advent of modern high-speed networks offers the possibility to transport VLBI data electronically with much higher data rates and lower costs to maximize the scientific potential of the observations. The electronic transmission of VLBI data (dubbed e-VLBI), however, presents a special challenge to the use of high-speed global networks. New protocols and strategies must be developed so that these applications become efficient high-bandwidth 'background' users, while not significantly impacting other normal network traffic. Because many high-speed networks, such as Abilene, are not heavily loaded on an average basis, there is considerable potential for applications of this class to productively use background capacity, promoting better science for fewer dollars. MIT Haystack Observatory, working jointly with the MIT Laboratory for Computer Science and MIT Lincoln Laboratory, proposes to undertake the following tasks to realize this potential: 1. Develop and test network protocols and algorithms that take advantage of the special characteristics of real-time scientific instrumentation applications to utilize low-priority bandwidth that is available in fair abundance on many networks. 2. Develop a system that can be deployed widely at low cost, that can be easily updated as algorithms improve, and that can track technology evolution to gain higher performance over time. 3. Disseminate the work to a broader community.