Research On The Key Technology Of Wide Bandwidth High-Resolution Digital Backedn For Pulsar Observation

Pulsars are highly magnetized,spinning and compact stars that are important in supernova explosion theory,plasma physics,general relativity,and stellar evolution research.One of the most significant components of the pulsar observation system is the pulsar digital backend.It converts analog pulsar signals received by the radio telescope into digital signals,and utilizes signal processing and storage in the digital domain.The performance requirements for the digital backend are continually growing as a result of the in-depth research of pulsars and the finding of more and more millisecond pulsars and high-dispersion pulsars.Real-time signal processing of pulsar signals with high time resolution and high frequency resolution become particularly necessary when broadband sampling is in use.This thesis studies the key technologies of high-bandwidth and high-resolution pulsar digital backends,to develop an Cascaded,Re-configurable Architecture Board(CRABoard)platform based on software defined radio technology.This platform has been used to finish the design of a ultra high frequency resolution pulsar digital back-end system and a real-time incoherent de-dispersion pulsar digital back-end system.The main research contents of this thesis are as follows:We investigate the effect of the earth’s revolution and rotation on the received pulsar signal period and present a pulsar period search technique.The method calculates the limit value of the period change considering the observation point at the earth’s equator and the limit of the earth’s unique orbital position.When observing pulsars on Earth,it is feasible to search for the precise signal period using the estimated limit value of the period variation and the step value of the period search.After the pulsar J0835-4510 was observed,the period value was corrected by 0.027‰ using the period search technique,generating a more accurate period superposition result.This research establishes the feasibility of using the period search technique to actual pulsar data.A high-performance integrated open hardware platform architecture is presented using the unified inter-module interface and cascading structure.The CRABoard digital back-end hardware platform,which uses FPGA as its core,is based on this architecture.The digital backend platform is separated into three hardware modules: analog-digital sampling module,realtime signal processing module,and control module.Any module may be simply and quickly upgraded using the designed unified inter-module interface without compromising the overall system’s functionality.Simultaneously,signal processing modules may be simply added to the layered structure,extending the hardware capabilities of the overall system.When many signal processing modules are overlaid,the signal processing module is configured using a daisy chain approach,allowing the control module to setup and control the whole system.We proposed the BIPC-FFT and IPBFD algorithms,which were integrated with the CORDIC method to achieve a high-performance parallel channelization technique.The BIPC-FFT algorithm combines complicated FFT with bit-reversed output to efficiently save store resources;the IPBFD algorithm decomposes the bit-reversed data when storage resources are restricted;the CORDIC algorithm is employed to conserves considerable storage space by keeping twiddle factors.The ultra-high frequency resolution digital back-end is developed on the CRABoard,which has two analog inputs,64 K frequency channles and 18.3k Hz frequency resolution.The combined parallel complex FFT method in the BIPC-FFT algorithm is used to channelize the dual-channel input signal,consuming 50% less hardware resources; the bit-reversed output minimizes resource consumption in the FFT core,and the CORDIC algorithm is employed.The twiddle factor table is implemented for FFT calculation with limited resources; the result of complex FFT calculation is solved using the IPBFD method; the power of the left and right polarized signals is split,and the BRAM resource needed by the calculation module is cut in half.The broadband real-time high-frequency resolution signal processing approach saves 80.1% of the BRAM resources and 50% of the slice resources.The BRAM resource that is ultimately utilised by the system consumes 92% of the entire chip,significantly increasing the usage rate of system resources.A pipelined real-time incoherent de-dispersion method(RT-PIDM)is presented.It’s based on the pulsar dispersion,and finishes the de-dispersion in pipeline in FPGA,and then uses the table to apply varying delays to incoming signals from different frequency channels,accumulating in the buffer for the associated time.Because the typical real-time incoherent dedispersion method must buffer all the data from all the frequency channels within the delay time,the buffer space is proportional to the product of the number of frequency channels and the delay time.The RT-PIDM algorithm does not need to cache the frequency channel data,but only the accumulation of the frequency channel addition.As a result,the algorithm’s buffer space is proportional to the delay time,which results in significant savings.It is well suited for implementing the algorithm in the FPGA,eliminating the requirement for GPU processing.This method is used as the heart of a real-time incoherent dedispersion system developed on the pulsar digital backend based on the CRABoard hardware platform.The system’s data output is significantly decreased,alleviating strain on following signal processing units.The CRABoard hardware platform,as well as the ultra-high frequency resolution pulsar digital back-end system and real-time incoherent dedispersion system built on top of it,have all been used to conduct actual observation experiments at the Shanghai Observatory’s Tianma Station and the Yunnan Observatory’s Kunming Station.The experimental findings demonstrate that a variety of constructed systems are capable of being applied in real-world scientific research.The findings of real-time dedispersion of J0835-4510 at Yunnan Astronomical Observatory’s Kunming Station were compared to those of PDFB(Pulsar Digital Filter Bank)dedispersion created by the Australian National Astronomical Observatory.