A Lossless Astronomical Data Compression Scheme with FPGA Acceleration

Abstract

Bottlenecks are arising from the movement of the massive radio observation data of the Five Hundred Meter Aperture Spherical Radio Telescope (FAST), a path-finder project of the Square Kilometer Array radio telescope (SKA) which is the biggest radio telescope being constructed. There are a few lossy compression approaches proposed by astronomers to reduce the size of original observation data of FAST before the data are transferred from a telescope site to a remote data center. However, these lossy methods would lead to missing critical hidden scientific information. To solve the bottlenecks without losing critical hidden scientific information, we propose a lossless compression scheme in this paper to exploit the internal redundancy in astronomical data. In our scheme, an efficient compression algorithm is proposed to optimize the data preprocessing by improving the organization and structure of the astronomical dataset. As the first step towards system-on-chip (SOC) application specific chip design, we evaluate the feasibility of hardware acceleration of our scheme by implementing a prototype of the compression scheme on FPGA platform. The compression scheme has achieved superior performance in compression ratio and power consumption compared to other state-of-the-art researches in astronomy domain.