Real-Time Signal Processor for Pulsar Studies

Journal of astrophysics and astronomy

P. S. Ramkumar²  A. A. Deshpande¹ 

[1] Raman Research Institute, C. V. Raman Avenue, Bangalore 560 080, India.$$;Present address: Intel Technology India Pvt Ltd, No. 65, 13th cross, III phase, JP Nagar, Bangalore 560 078.$$

关键词: Stars: neutron;    pulsars;    interstellar medium: dispersion;    Faraday rotation;    telescope: GMRT;    instrumentation.;   

DOI  :  

学科分类：天文学(综合)

来源: Indian Academy of Sciences

【 摘 要 】

This paper describes the design, tests and preliminary results of a real-time parallel signal processor built to aid a wide variety of pulsar observations. The signal processor reduces the distortions caused by the effects of dispersion, Faraday rotation, doppler acceleration and parallactic angle variations, at a sustained data rate of 32 Msamples/sec. It also folds the pulses coherently over the period and integrates adjacent samples in time and frequency to enhance the signal-to-noise ratio. The resulting data are recorded for further off-line analysis of the characteristics of pulsars and the intervening medium. The signal processing for analysis of pulsar signals is quite complex, imposing the need for a high computational throughput, typically of the order of a Giga operations per second (GOPS). Conventionally, the high computational demand restricts the flexibility to handle only a few types of pulsar observations. This instrument is designed to handle a wide variety of Pulsar observations with the Giant Metre Wave Radio Telescope (GMRT), and is flexible enough to be used in many other high-speed, signal processing applications. The technology used includes field-programmable-gate-array(FPGA) based data/code routing interfaces, PC-AT based control, diagnostics and data acquisition, digital signal processor (DSP) chip based parallel processing nodes and C language based control software and DSP-assembly programs for signal processing. The architecture and the software implementation of the parallel processor are fine-tuned to realize about 60 MOPS per DSP node and a multiple-instruction-multiple-data (MIMD) capability.

【 授权许可】

Unknown