云台40米射电望远镜脉冲星后端实时pipeline关键技术研究

脉冲星的宽带观测面临着海量的实时数据采集和处理的挑战，而当前国内的高性能的脉冲星观测后端数据处理系统主要还是由国外引进。本项目针对中科院云南天文台40米射电望远镜宽带脉冲星后端处理的实际需要，研究用户空间态的下海量实时数据采集，GPU/CPU异构并行的海量数据实时处理，脉冲星数据管线构建，以及基于轻量级容器Docker技术的敏捷化部署的关键技术与方法。在40米射电望远镜上实现一套自主研发的高性能的数字观测后端，为脉冲星到达时间、单脉冲等观测提供高性能实时相干消色散处理。本项目的关键技术，可用于快速构建脉冲星实时数据处理管线，其数据采集、解码、消色散、持久化、归档和可视化等可重用组件，为非相关消色散、谱线等观测模式的扩展提供支撑，同时也对云台筹建中的景东120米射电望远镜的后端系统的设计有着较好的参考价值。

The wide-band observations of pulsars face the challenge of massive real-time data acquisition and processing. Currently high-performance pulsar observation back-end data processing systems are mainly imported from abroad. This project aims at the actual requirements of the processing of pulsar backend at the 40-meter radio telescope of the Yunnan Observatory of the Chinese Academy of Sciences. Based on the previous work, it focuses on massive real-time data acquisition based on the user-space IO in a high IO network environment, massive real-time data handling based on GPU/CPU heterogeneous parallelism, key methods for the construction of pulsar data pipelines， and the astronomical real-time processing cluster agile construction and deployment based on lightweight container Docker technology. A self-developed high-performance observation back-end will be implemented on the 40-meter radio telescope, providing high-performance real-time coherent de-dispersion processing for time of arrival, single-pulse observations and other observations. The key technology of this project can be used to rapidly construct a real-time data processing pipeline for pulsar observations, and its reusable components such as data acquisition, decoding, de-dispersion, persistence, archiving, and visualization can be used as incoherent de-dispersion, spectral lines modes. It also has a good reference value for the design of the back-end systems of JingDong 120-m radio telescopes in the future.