KMS of Xinjiang Astronomical Observatory, CAS
| 射电望远镜相控阵接收机信号处理关键技术研究 | |
| Alternative Title | Research on Key Techniques of Signal Processing for Radio Telescope Phased Array Feed (PAF) Receiver |
裴鑫
| |
| Subtype | 博士 |
| Thesis Advisor | 王娜 |
| 2022-05-25 | |
| Degree Grantor | 中国科学院大学 |
| Place of Conferral | 北京 |
| Degree Name | 理学博士 |
| Degree Discipline | 天文技术与方法 |
| Keyword | 射电天文，数字接收机，相控阵馈源，数字波束合成，射频直接采样 |
| Abstract | 大型射电望远镜一项突出的优势是通过增加口径来提高观测灵敏度，以探测更加暗弱的天体，常用来进行未知天体的搜寻发现。然而，大型望远镜的观测视场却随着口径的增加而减小，导致单位时间内观测天区的面积减小，巡天效率下降。相控阵馈源（PAF）是近些年在射电天文得到大力发展的一种多波束接收机技术，它可以将多个馈电单元安装于望远镜焦平面，通过电子扫描来形成多个同步波束，从而增大望远镜的观测视场，提高巡天效率。同时，这些密集交叠的波束还可以形成连续的天空覆盖，通过动态波束合成可以实现多种灵活的观测模式。PAF接收机是大型射电望远镜的巡天利器。 下一代射电天文PAF阵元数量达到上百个，信号带宽高达GHz，PAF前端设计复杂度高，信号处理系统开发挑战大。本文引入数字化接收机设计思路，采用新一代高集成度、高性能射频片上系统（RFSoC）技术，靠近PAF阵元数字化，降低模拟前端复杂度，减小空间和重量，提高信号采集保真度，满足馈源阵列紧凑化设计要求。本研究面向下一代大规模、超宽带PAF接收机信号处理关键技术，开展了数字化前端设计、数字波束合成网络设计以及相关仿真实验，论文完成的主要工作如下： （1）对RFSoC平台的采样特性、功耗、电磁辐射特性、高速传输端口等进行了测试，评估了RFSoC技术应用到射电天文PAF数字接收机设计的可行性。测试结果表明RFSoC集成式设计较分立式信号采集功耗降低一半以上、电磁辐射更弱，RFSoC信号采集和数据卸载能力强、信号采集质量高，非常适合大规模、宽带PAF信号的采集和处理。 （2）开发了基于RFSoC的数字化PAF前端，在CASPER开发工具流的辅助下，设计了最高采样速率4.096 GSPS、12 bit精度、8通道信号采集和预处理固件程序，根据波束合成计算特征开发了信道化、多级数据重排和格式化等模块，将多通道宽带信号变换组合为多条窄带数据流通过多条高速网络链路输出，实现了PAF信号的超高速采集与多路径交叉分发。开发了基于GPU服务器集群的实时数字波束合成后端，采用高可获取共享管道引擎框架和交叉连接波束合成网络拓扑，设计了HRBF_HASHPIPE数字波束合成算法，并采用锁页内存、批处理、非统一内存访问绑定和获取原始套接字等方法对算法和计算架构进行了优化，实现了超高速数据流的多线程实时并行处理，单台服务器搭载4块RTX 3090 GPU运行4个实例下的信号处理带宽达到256 MHz，波束合成参数可动态调整。 （3）建立了RFSoC+GPU混合架构的PAF信号采集与实时处理仿真实验平台。采用软件模拟产生测试数据对GPU波束合成算法的准确性进行了验证；对实验平台的信号采集同步性和数据传输完整性进行了测试，并采用2块RFSoC平台直接采集Vivaldi天线阵列射频信号，开展了不同阵元处理规模和波束合成数量的仿真实验。实验验证了算法的有效性和实验平台的准确、高效，建立了灵活度高、可扩展性好的混合架构集群波束合成网络计算模型，为下一代PAF的多通道、超宽带信号采集和海量数据实时处理提供了技术参考。 （4）针对新疆110米口径全可动射电望远镜（QTT）规划的20 cm PAF接收机参数指标，规划了基于RFSoC的数字化PAF前端设计方案，给出了三种配置下的基于RFSoC+GPU混合架构的分布式PAF信号处理与波束合成网络设计，仿真分析了不同波束合成算法的特征和面对干扰的自适应性，仿真分析了PAF阵元间距、阵元排列形状、合成波束数量、子阵阵元数量等参数对波束合成计算复杂度和观测视场的影响，为QTT波束合成系统的设计提供参考。 |
| Other Abstract | One of the outstanding advantages of large radio telescopes is to increase the observation sensitivity by increasing the aperture to detect fainter celestial bodies. It is often used to search and discover unknown celestial bodies. However, the observation Field of View (FoV) of large telescopes decreases with the increase of the aperture, resulting in a decrease in the area of the observation sky per unit time and low sky survey efficiency. Phased Array Feed (PAF) is a multi-beam receiver technology that has been widely developed in radio astronomy in recent years. Multiple feed units can be installed on the focal plane of the telescope and form multiple synchronized beams through electronic scanning. Which can increase the observation FoV of the telescope and improve the efficiency of sky surveying. Meanwhile, these densely overlapping beams can form continuous sky coverage. Furthermore, through dynamic beamforming, a variety of flexible observation modes can be realized. In one word, the PAF receiver is a powerful tool for sky surveying by large radio telescopes. The number of next-generation radio astronomy PAF array elements reaches hundreds, and the signal bandwidth is up to GHz, which increases the complexity of PAF front-end design and poses great challenges to the development of signal processing systems. The thesis introduces the design concept of the digital receiver and attempts to use a new generation of high-integration, high-performance Radio Frequency System-on-Chip (RFSoC) technology to digitize close to the feed array, reduce the complexity and analog components of the front-end, reduce space and weight, improve the fidelity of the signals, and meet the compact design requirements of the feed array. Aiming at the signal processing key technologies of the next generation of large-scale, ultra-wideband PAF receivers, the thesis carried out digital front-end design, digital beamforming network design and related simulation experiments. The main work completed by this thesis is as follows: (1) The thesis tests the sampling characteristics, power consumption, electromagnetic radiation characteristics, high-speed transmission ports, etc. of the RFSoC board, and evaluates the feasibility of applying RFSoC technology to the design of radio astronomy PAF digital receivers. The results show that the RFSoC integrated design reduces the power consumption by more than half and the electromagnetic radiation is weaker compared to the discrete signal acquisition. The RFSoC has strong signal acquisition and data offloading capabilities and high signal acquisition quality, making it very suitable for large-scale, broadband PAF signal acquisition and processing. (2) The digital PAF front-end is designed based on the RFSoC board. With the assistance of the CASPER development toolkit, an 8-channel input signal acquisition and pre-processing firmware is developed, with a maximum sampling rate of 4.096 GSPS and a sampling accuracy of 12 bit. According to the characteristics of beamforming calculation, the channelization, reordering and formatting modules are designed to transform and combine multi-channel broadband signals into multiple narrow-band data streams, which are output through multiple high-speed network links, realizing ultra-high-speed acquisition and multi-path cross-distribution of PAF signals. The digital beamforming backend is implemented on GPU servers. Based on the high-availability shared pipeline engine framework and the cross-connected beamforming network topology, a high-throughput real-time beamforming algorithm (HRBF_HASHPIPE) is designed. The algorithm and computing architecture are optimized by using methods such as page-locking memory, batch processing, NUMA binding, and raw socket acquisition, realizing multi-threaded real-time parallel processing of ultra-high-speed data streams. A single server equipped with four RTX 3090 GPUs can reach 256 MHz of signal processing bandwidth under four instances, and the beamforming parameters can be dynamically adjusted. (3) The thesis establishes a RFSoC+GPU hybrid architecture PAF signal acquisition and processing experimental platform. The thesis uses software to generate test data to verify the accuracy of the digital beamforming algorithm. Then, the thesis conducts signal acquisition synchronization and data transmission integrity tests on the hybrid architecture experiment platform, and carries out simulation experiments with different processing array elements and an adjustable number of formed beams, two RFSoC platforms are used to directly collect the RF signals of the Vivaldi antenna array. The results show that the algorithm is effective and the experimental platform is accurate and efficient. The experiment establishes a hybrid architecture cluster beamforming network computing model with high flexibility and good scalability, which provides a technical reference for multi-channel, ultra-wideband signal acquisition and high-speed data real-time processing of next-generation PAFs. (4) Aiming at the 20 cm PAF receiver planned by the Qi Tai radio Telescope (QTT) project, the thesis presents the digital PAF front-end design scheme based on RFSoC, and demonstrates distributed QTT PAF signal processing and beamforming network designs based on RFSoC+GPU hybrid architecture in three configurations. The thesis simulates and analyzes the characteristics of different beamforming algorithms and their adaptability to suppress interference. The effects of parameters such as the distance between PAF array elements, the arrangement shape of array elements, the number of formed beams, and the number of sub-array elements on the computational complexity of beamforming and the field of view are simulated and analyzed. These works will provide design references for the development of the QTT beamforming system. |
| Pages | 121 |
| Language | 中文 |
| Document Type | 学位论文 |
| Identifier | http://ir.xao.ac.cn/handle/45760611-7/5162 |
| Collection | 研究生学位论文 |
| Affiliation | 中国科学院新疆天文台 |
| First Author Affilication | Xinjiang Astronomical Observatory, Chinese Academy of Sciences |
| Recommended Citation GB/T 7714 | 裴鑫. 射电望远镜相控阵接收机信号处理关键技术研究[D]. 北京. 中国科学院大学,2022. |
| Files in This Item: | ||||||
| File Name/Size | DocType | Version | Access | License | ||
| 射电望远镜相控阵接收机信号处理关键技术研（23820KB） | 学位论文 | 暂不开放 | CC BY-NC-SA | Application Full Text | ||
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.
Edit Comment