基于缓存区搬移策略的 GPU高效信道化方法

An efficient channelization method on GPU with buffer shift strategy

摘要

信道化处理是现代电子战数字系统中的首要任务。基于CPU的信道化处理以多相滤波为基础，使用分段卷积的方式保证信道化结果的相位连续。然而，随着数据量的增大，该方法无法满足实时处理的要求，研究如何基于GPU实现高性能信道化处理是目前急需解决的问题。首先分析了传统分段卷积方法在GPU架构上的低效性，随后结合GPU架构特点提出了一种缓存空间需求更低、计算量更低、逻辑控制更方便的缓存区搬移策略来保证信道化结果的相位连续性。此外，分析了在GPU架构下基于多相滤波和直接滤波的两种多级滤波方式，说明了GPU架构下采用直接滤波方式的优越性。仿真实验表明所提方法能正确保证信道化结果的相位连续性，给出了GPU和CPU下实现直接滤波的加速比，直观说明了基于GPU的高效信道化方法带来运算效率的巨大提升，同时GPU架构下直接滤波快于多相滤波。所提的基于缓存区搬移策略的GPU高效信道化方法在处理速度和数据相位连续性上具有显著优势，尤其适用于大规模数据的实时处理应用。

Abstract

Channelization processing is the first task in modern electronic warfare digital systems. CPU-based channelization processing is based on polyphase filtering and segmented convolution to ensure phase continuity of the channelized results. However, as the increasing of data amount, it cannot meet the requirements of real-time processing. The study on how to implement high-performance channelization processing based on GPU is currently an urgent issue that needs to be addressed. The inefficiency of traditional segmented convolution methods on GPU architecture is first analyzed, then a buffer shift strategy is proposed to ensure the phase continuity of channelization results with lower buffer space, reduced computational overhead and simplified logic control. Additionally, the analysis of two multistage filtering approaches, i.e., polyphase filtering and direct filtering, demonstrates the superiority of direct filtering on GPU architecture. Simulation experiments show that the proposed method correctly ensures the phase continuity of the channelization results and confirms that direct filtering is faster than polyphase filtering on GPU architecture. The acceleration ratios of direct filtering using GPU and CPU are provided to illustrate the substantial improvement in computational efficiency by the proposed strategy. In a word, the proposed buffer shift strategy-based GPU-efficient channelization method has significant advantages in processing speed and data phase continuity and is particularly suitable for real-time processing of large-scale data.

关键词

信道化处理 / GPU / 相位连续性 / 多相滤波 / 直接滤波

Key words

channelization processing / GPU / phase continuity / polyphase filtering / direct filtering

参考文献

[1] HAN W W, LIU Z H, XU Y F, et al. Photonic RF channelization applications of microcombs[J]. IEEE Journal of Selected Topics in Quantum Electronics, 2024, 30(5): 7600417.

[2] 杨静, 吕幼新 . 高效数字信道化IFM接收机的研究[J]. 电子科技大学学报, 2005, 34(4): 444-447.

[3] YANG J, LYU Y X , Efficient digital channelized IFM receiver research[J]. Journal of University of Electronic Science and Technology of China, 2005, 34(4): 444-447.

[4] KIM S, PLISHKER S, BHATTACHARYYA S. An efficient GPU implementation of an arbitrary resampling polyphase channelizer[C]// 2013 Conference on Design and Architectures for Signal and Image Processing. Cagliar: [s.n.], 2013: 231-238.

[5] ADHINARAYANAN V, FENG W C. Wideband channelization for software—defined radio via mobile graphics processors[C]// Proceedings of the International Conference on Parallel and Distributed Systems. New York: IEEE, 2013: 86-93.

[6] ZHANG W X, YAO Y S, ZHAO Z K, et al. Design and FPGA implementation of a novel efficient FRM—based channelized receiver structure[J]. IEEE Access, 2019, 7: 114778-114787.

[7] MAHESH R, VINOD A P. Reconfigurable frequency response masking filters for software radio channelization[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2008, 55(3): 274-278.

[8] AMBEDE A, SHREEJITH S, VINOD A P, et al. Design and realization of variable digital filters for software—defined radio channelizers using an improved coefficient decimation method[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2016, 63(1): 59-63.

[9] 匡宏印 . 一种基于时间−频率特征分析的自适应信道化技术[J]. 电讯技术, 2022, 62(3): 367-372.

[10] KUANG H Y. An adaptive channelized technology based on time—frequency analysis[J]. Telecommunication Engineering, 2022, 62(3): 367-372.

[11] XU S C, LIU G M, GAO M G. Design and implementation of digital channelized receiver in multi—FPGA[C]// Proceedings of the 1st International Conference on Information Science and Engineering. New York: IEEE, 2009: 178-181.

[12] DEL MUNDO C, ADHINARAYANAN V, FENG W C. Accelerating fast Fourier transform for wideband channelization[C]// Proceedings of the IEEE International Conference on Communications. New York: IEEE, 2013: 4776-4780.

[13] MAHESH R, VINOD A P. Low complexity flexible filter banks for uniform and non—uniform channelisation in software radios using coefficient decimation[J]. IET Circuits, Devices & Systems, 2011, 5(3): 232-242.

[14] 徐庆征, 彭丽 . 基于MATLAB的分段卷积计算[J]. 苏州科技学院学报: 工程技术版, 2006, 19(2): 88-90.

[15] XU Q Z, PENG L. Calculation of partition convolution based on the MATLAB[J]. Journal of University of Science and Technology of Suzhou: Engineering and Technology, 2006, 19(2): 88-90.