宽视场成像网格化算法中w-plane最优经验值研究
|
摘要
射电干涉阵列宽视场成像网格化处理过程中必须考虑w项的影响。w-projection和w-stacking是两种重要的宽视场成像网格化处理算法,w-plane参数是算法中影响计算速度和成图质量的一个重要因素。研究了w-projection和w-stacking两种网格化算法,利用SKA-1低频阵台站数据和ASKAP软件包进行模拟观测,对两种算法在不同w-plane参数取值情况下的成图速度和成图质量进行了定量分析对比。结果进一步表明,w-plane是性能改善的重要参数。针对w-projection算法,w-plane取值应比一般给定的经验值大才能得到较好的成像效果。w-stacking算法虽然有很大的速度优势,但算法实现中w-plane的影响更为显著,给出了推荐的w-plane取值。本文的工作是大视场成像算法的基础性研究工作,对未来平方千米阵列科学数据处理中的管线设计有较好的参考价值。
Astronomy data processing platform is facing unprecedented big data challenge. SKA(square kilometer array) as the biggest radio synthesis telescope, its generated data rate of averaged visibilities will be hundred gigabit per second. SKA-SDP handles big data and extremely large computing cost by data processing pipeline. Gridding algorithm will take a large proportion of compute load. How to save power cost and improve the computational efficiency, on the prerequisite of guaranteed imaging quality, need to be well-considered. W-projection and w-stacking are two gridding algorithms possibly to be used in SKA-SDP; The paper presents a study on the effect of w-plane in those algorithms. Based on SKA continuum imaging pipeline and simulation observation, we analyze the effect of w-plane to the imaging results. Our experimental results show that w-plane is a crucial parameter for performance promotion, and w-plane should be bigger than experimental value in w-projection algorithm. Compared with w-projection algorithm, w-stacking has advantage in run time, yet w-plane has great effect on w-stacking algorithm. The presented paper is a basic work for study on large field imaging, and a new w-plane for w-projection and w-stacking algorithm is proposed. Our study can serve as a reference for designs of SKA data processing pipeline.
Astronomy data processing platform is facing unprecedented big data challenge. SKA(square kilometer array) as the biggest radio synthesis telescope, its generated data rate of averaged visibilities will be hundred gigabit per second. SKA-SDP handles big data and extremely large computing cost by data processing pipeline. Gridding algorithm will take a large proportion of compute load. How to save power cost and improve the computational efficiency, on the prerequisite of guaranteed imaging quality, need to be well-considered. W-projection and w-stacking are two gridding algorithms possibly to be used in SKA-SDP; The paper presents a study on the effect of w-plane in those algorithms. Based on SKA continuum imaging pipeline and simulation observation, we analyze the effect of w-plane to the imaging results. Our experimental results show that w-plane is a crucial parameter for performance promotion, and w-plane should be bigger than experimental value in w-projection algorithm. Compared with w-projection algorithm, w-stacking has advantage in run time, yet w-plane has great effect on w-stacking algorithm. The presented paper is a basic work for study on large field imaging, and a new w-plane for w-projection and w-stacking algorithm is proposed. Our study can serve as a reference for designs of SKA data processing pipeline.
引文
[1] DEWDNEY P E, HALL P J, SCHILIZZI R T, et al. The square kilometre array[J]. Proceedings of the IEEE, 2009, 97: 1482-1496.
[2] NIKOLIC B, ALEXANDER P. PDR. 01 SDP architecture[EB/OL]. 2015[2018-09-13]. http://www.astron.nl/～broekema/papers/SDP-PDR/PDR01%20System%20Architecture.pdf.
[3] CORNWELL T J, PERLEY R A. Radio-interferometric imaging of very large fields-the problem of non-coplanar arrays[J]. Astronomy & Astrophysics, 1992, 261(1): 353-364.
[4] TAYLOR G B, CARILLI C L, PERLEY R A. Synthesis Imaging in Radio Astronomy II[C]// ASP Conference Series. 1999: 383.
[5] CORNWELL T J, GOLAP K, BHATNAGAR S. The noncoplanar baselines effect in radio interferometry: the w-projection algorithm[J]. IEEE Journal of Selected Topics in Signal Processing, 2008, 2(5): 647-657.
[6] HUMPHREYS B, CORNWELL T J. Analysis of convolutional resampling algorithm performance[EB/OL].[2018-09-13]. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.190.6087&rep=rep1&type=pdf.
[7] 劳保强, 安涛, 陈骁, 等. 低频射电望远镜阵列宽视场成像技术研究[J]. 天文学报, 2017, 58(5): 108-129.
[8] DEBOER D R, GOUGH R G, BUNTON J D, et al. Australian SKA Pathfinder: a high-dynamic range wide-field of view survey telescope[J]. Proceedings of the IEEE, 2009, 97(8): 1507-1521.
[9] Cornwell T J, HUMPHREYS B, LENC E, et al. ASKAP science processing[EB/OL]. 2016[2018-09-13]. http://www.atnf.csiro.au/projects/askap/ASKAP-SW-0020.pdf.
[10] CHAPMAN J M. CASDA: the CSIRO ASKAP science data archive: requirements and use cases [EB/OL]. 2013[2018-09-13]. http://www.atnf.csiro.au/management/atuc/2013dec/docs/ASKAP_SW_0017_v0.8_fordistribution.pdf.
[11] GOODRICK L. Image reconstruction in radio astronomy with non-coplanar synthesis arrays[D]. Matieland: Electrical and Electronic Engineering University of Stellenbosch, 2015.
[2] NIKOLIC B, ALEXANDER P. PDR. 01 SDP architecture[EB/OL]. 2015[2018-09-13]. http://www.astron.nl/～broekema/papers/SDP-PDR/PDR01%20System%20Architecture.pdf.
[3] CORNWELL T J, PERLEY R A. Radio-interferometric imaging of very large fields-the problem of non-coplanar arrays[J]. Astronomy & Astrophysics, 1992, 261(1): 353-364.
[4] TAYLOR G B, CARILLI C L, PERLEY R A. Synthesis Imaging in Radio Astronomy II[C]// ASP Conference Series. 1999: 383.
[5] CORNWELL T J, GOLAP K, BHATNAGAR S. The noncoplanar baselines effect in radio interferometry: the w-projection algorithm[J]. IEEE Journal of Selected Topics in Signal Processing, 2008, 2(5): 647-657.
[6] HUMPHREYS B, CORNWELL T J. Analysis of convolutional resampling algorithm performance[EB/OL].[2018-09-13]. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.190.6087&rep=rep1&type=pdf.
[7] 劳保强, 安涛, 陈骁, 等. 低频射电望远镜阵列宽视场成像技术研究[J]. 天文学报, 2017, 58(5): 108-129.
[8] DEBOER D R, GOUGH R G, BUNTON J D, et al. Australian SKA Pathfinder: a high-dynamic range wide-field of view survey telescope[J]. Proceedings of the IEEE, 2009, 97(8): 1507-1521.
[9] Cornwell T J, HUMPHREYS B, LENC E, et al. ASKAP science processing[EB/OL]. 2016[2018-09-13]. http://www.atnf.csiro.au/projects/askap/ASKAP-SW-0020.pdf.
[10] CHAPMAN J M. CASDA: the CSIRO ASKAP science data archive: requirements and use cases [EB/OL]. 2013[2018-09-13]. http://www.atnf.csiro.au/management/atuc/2013dec/docs/ASKAP_SW_0017_v0.8_fordistribution.pdf.
[11] GOODRICK L. Image reconstruction in radio astronomy with non-coplanar synthesis arrays[D]. Matieland: Electrical and Electronic Engineering University of Stellenbosch, 2015.