一種在線時間序列預測的核自適應(yīng)濾波器向量處理器
doi: 10.11999/JEIT150157 cstr: 32379.14.JEIT150157
基金項目:
國家自然科學基金(61571160/F011305)，中央高?；究蒲袠I(yè)務(wù)費專項資金資助(HIT.NSRIF.201615)
A Kernel Adaptive Filter Vector Processor for Online Time Series Prediction
Funds:
The National Natural Science Foundation of China (61571160/F011305), Fundamental Research Funds for the Central Universities (HIT.NSRIF.201615)
-
摘要: 針對信息物理融合系統(tǒng)中的在線時間序列預測問題，該文選擇計算復雜度低且具有自適應(yīng)特點的核自適應(yīng)濾波器(Kernel Adaptive Filter, KAF)方法與FPGA計算系統(tǒng)相結(jié)合，提出一種基于FPGA的KAF向量處理器解決思路。通過多路并行、多級流水線技術(shù)提高了處理器的計算速度，降低了功耗和計算延遲，并采用微碼編程提高了設(shè)計的通用性和可擴展性。該文基于該向量處理器實現(xiàn)了經(jīng)典的KAF方法，實驗表明，在滿足計算精度要求的前提下，該向量處理器與CPU相比，最高可獲得22倍計算速度提升，功耗降為1/139，計算延遲降為1/26。
-
關(guān)鍵詞:
- 核自適應(yīng)濾波器(KAF) /
- 現(xiàn)場可編程邏輯門陣列(FPGA) /
- 向量處理器 /
- 微碼
Abstract: To address the online time series prediction problem in CPS (Cyber-Physical System) system, both KAF (Kernel Adaptive Filter) with low computation complexity and adaptive characteristic and FPGA computing system are employed. A novel FPGA implementation of vector processor targeting KAF algorithm is proposed. The parallelized datapath and multi-stage pipeline are introduced to enhance the performance and reduce the power consumption and latency. The microcoding technology is further employed to improve the reusability and extensibility. The classical KAF algorithms are implemented based on the vector processor. Experiments results show that the proposed vector processor improves the execution speed by factors of 22, the power consumption decrease to 1/139, while the latency decrease to 1/26 compared with a CPU, on the condition that the precision meets the requirement. -
王中杰, 謝璐璐. 信息物理融合系統(tǒng)研究綜述[J]. 自動化學報, 2011, 37(10): 1157-1166. WANG Zhongjie and XIE Lulu. Cyber-physical system: a survey[J]. Acta Automatica Sinica, 2011, 37(10): 1157-1166. 周建寶, 王少軍, 馬麗萍, 等. 可重構(gòu)衛(wèi)星鋰離子電池剩余壽命預測系統(tǒng)研究[J]. 儀器儀表學報, 2013, 34(9): 2034-2044. ZHOU Jianbao, WANG Shaojun, MA Li-ping, et al. Study on the reconfigurable remaining useful life estimation system for satellite lithium-ion battery[J]. Chinese Journal of Scientific Instrument, 2013, 34(9): 2034-2044. 王少軍. 時間序列預測的可重構(gòu)計算研究[D]. [博士論文], 哈爾濱工業(yè)大學, 2012. WANG Shaojun. Research on reconfigurable computing for time series forecasting[D]. [Ph.D. dissertation], Harbin Institute of Technology, 2012. 曹葵康. 支持向量機加速方法及應(yīng)用研究[D]. [博士論文], 浙江大學, 2010. CAO Kuikang. Acceleration and application of support vector machines[D]. [Ph.D. dissertation], Zhejiang University, 2010. 江潔, 凌思睿. 一種投票式并行RANSAC算法及其FPGA實現(xiàn)[J]. 電子與信息學報, 2014, 36(5): 1145-1150. doi: 10.3724/ SP.J.1146.2013.00962. JIANG Jie and LING Sirui. Parallel voting RANSAC and its implementation on FPGA[J]. Journal of Electronics Information Technology, 2014, 36(5): 1145-1150. doi: 10. 3724/SP.J.1146.2013.00962. 蘭亞柱, 楊海鋼, 林郁. 動態(tài)自適應(yīng)低密度奇偶校驗碼譯碼器的FPGA實現(xiàn)[J]. 電子與信息學報, 2015, 37(8): 1937-1943. doi: 10.11999/JEIT141609. LAN Yazhu, YANG Haigang, and LIN Yu. Design of dynamic adaptive LDPC decoder based on FPGA[J]. Journal of Electronics Information Technology, 2015, 37(8): 1937-1943. doi: 10.11999/JEIT141609. PAPADONIKOLAKIS M. A scalable FPGA architecture for non-linear svm training[C]. Proceeding of 2008 International Conference on Field Programmable Technology, Taipei, China, 2008: 337-340. ANGUITA D, CARLINO L, GHIO A, et al. A FPGA core generator for embedded classification systems[J]. Journal of Circuits, Systems and Computers, 2011, 20(2): 263-282. MAJUMDAR A, CADAMBI S, BECCHI M, et al. A massively parallel, energy efficient programmable accelerator for learning and classification[J]. ACM Transactions on Architecture and Code Optimization, 2012, 9(1): 6:1-6:30. KOZYRAKIS C and PATTERSON D. Vector vs. superscalar and vliw architectures for embedded multimedia benchmarks[C]. Proceeding of 35th Annual IEEE/ACM International Symposium on Microarchitecture, Califonia, USA, 2002: 283-293. YIANNACOURAS P, STEFFAN J G, and ROSE J. Portable, flexible, and scalable soft vector processors[J]. IEEE Transactions on Very Large Scale Integration Systems, 2012, 20(8): 1429-1442. YU J, EAGLESTON C, CHOU C H Y, et al. Vector processing as a soft processor accelerator[J]. ACM Transactions on Reconfigurable Technology and Systems, 2009, 2(2): 12:1-12:34. VAN VAERENBERGH S, VIA J, and SANTAMARIA I. A sliding-window kernel RLS algorithm and its application to nonlinear channel identification[C]. 2006 IEEE International Conference on Acoustics, Speech and Signal Processing, Toulouse, France, 2006: 789-792. PANG Yeyong, WANG Shaojun, PENG Yu, et al. A low latency kernel recursive least squares processor using FPGA technology[C]. 2013 International Conference on Field- Programmable Technology (FPT), Kyoto, Japan, 2013: 144-151. VAN VAERENBERGH S, SANTAMARIA I, WEIFENG L, et al. Fixed- budget kernel recursive least-squares[C]. 2010 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), Dalas, USA, 2010: 1882-1885. RICHARD C, BERMUDEZ J C M, and HONEINE P. Online prediction of time series data with kernels[J]. IEEE Transactions on Signal Processing, 2009, 57(3): 1058-1067. KRUIF B J and VRIES T J A. Pruning error minimization in least squares support vector machines[J]. IEEE Transactions on Neural Networks, 2003, 14(3): 696-702. LEONG P H W and LEUNG I K H. A microcoded elliptic curve processor using FPGA technology[J]. IEEE Transactions on VLSI Systems, 2002, 10(5): 550-559. CHAU T C P, KUREK M, TARGETT J S, et al. SMCGen: Generating reconfigurable design for sequential Monte Carlo applications[C]. Proceeding of 22nd IEEE Symposium on Field-Programmable Custom Computing Machines (FCCM), Boston, USA, 2014: 141-148. VAN VAERENBERGH S and SANTAMARIA I. A comparative study of kernel adaptive filtering algorithms[C]. Proceedings of 2013 IEEE Digital Signal Processing and Signal Processing Education Meeting, California, USA, 2013: 181-186. -
計量
- 文章訪問數(shù): 1735
- HTML全文瀏覽量: 166
- PDF下載量: 726
- 被引次數(shù): 0
下載:
