MRI: Development of a Cloud-Computing Platform for Interferometric Processing

点击翻译按钮获取中文摘要

The technique of Very Long Baseline Interferometry (VLBI) links together radio dishes around the globe to form Earth-sized telescopes. Each telescope in the array, synchronized by a network of atomic clocks, observes the same cosmic target at the same moment, recording data onto banks of hard-disk drives. Currently, these disks are sent to a purpose-built computer cluster for processing and imaging. But the sensitivity required for the most cutting edge work in this field has led to recording rates and data volumes that now exceed the capacity of in-house computing clusters. This project will include a partnership with Google to develop infrastructure and software that will shift VLBI processing to the cloud, where enormous virtual machines can be nimbly configured to process scientific data at speeds that easily keep pace with advancing recording rates. This project will revolutionize the practice of VLBI, and provide an extensible means of increasing array bandwidths and spectral resolution for a broad range of science. At the absolute limits of the VLBI technique, a millimeter wavelength global array will image nearby supermassive black holes as part of the Event Horizon Telescope (EHT). This project will also include educational participation and early career development opportunities in instrumentation and astronomy research.This project will develop a scalable, cloud-based computational system for VLBI. This project is motivated by the computational needs of extending EHT observations to short wavelengths; however, it will develop a methodology that is broadly applicable to the current and future needs of VLBI observations. The project will develop a hardware/software approach in conjunction with commercial partner, Google. This approach will include a new recording system for VLBI using customized commercial appliances for the transfer of data from the telescope to the cloud, and modifications to existing correlator code to make it suitable for execution on a parallelized, cloud-based platform. This project will contribute to broadening participation in science through student research opportunities made in conjunction with the Urban Massachusetts Louis Stokes Alliance for Minority Participation Program and the Akamai Workforce Initiative.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.