计算机科学 ›› 2020, Vol. 47 ›› Issue (6): 242-246.doi: 10.11896/jsjkx.190500080
林云, 黄桢航, 高凡
LIN Yun, HUANG Zhen-hang, GAO Fan
摘要: 目前大多数分布式估计算法以最小均方误差准则作为代价函数,在脉冲噪声下性能恶化乃至发散。扩散式仿射投影符号算法(Diffusion Affine Projection Sign Algorithm,DAPSA)以L1范数为代价函数,在脉冲噪声环境中具有良好的鲁棒性,并且具有较快的收敛速度。然而,固定步长的DAPSA在保持较大的初始收敛速度和较低的稳态误差之间存在矛盾。为降低非高斯噪声环境下DAPSA的稳态误差,同时仍保持较快的初始收敛速度,文中提出了一种扩散式最大相关熵准则变步长仿射投影符号算法(Diffusion Maximum Correntropy Criterion Variable StepSize Affine Projection Sign Algorithm,DMCCVSS-APSA)。首先,该算法利用改进的卡方核作为核函数,自适应更新算法每次迭代过程中的步长取值,在取得较快初始收敛速度的同时可有效降低稳态误差;然后,提出了一种基于系统先验误差的自适应动态范围方法,以进一步降低稳态误差;最后,通过改进卡方核与改进高斯核函数的对比实验,DMCCVSS-APSA与其他分布式算法的对比实验、不同脉冲噪声环境下DMCCVSS-APSA和DAPSA的对比实验,验证了所提算法的性能表现。仿真结果表明,DMCCVSS-APSA与对比算法相比表现良好,在相似的初始收敛速度下稳态误差降低了5dB以上。实验数据充分说明,在固定步长的DAPSA的基础上提出的变步长方法和自适应动态范围方法,具有对脉冲噪声的强鲁棒性的同时,能有效降低稳态误差,提升了分布式仿射投影类算法的性能表现。最后指出所提算法在ATC式联合方式和最优灵敏度的取值上需要进一步的研究。
中图分类号:
[1]SAYED A H.Adaptive Networks[J].Proceedings of the IEEE,2014,102(4):460-497. [2]CHEN J,RICHARD C,TING S K,et al.Multitask Learning Over Adaptive Networks With Grouping Strategies[M]//Cooperative and Graph Signal Processing.Academic Press,2018:107-129. [3] HU J P,ZHENG W X.Adaptive tracking control of leader-follower systems with unknown dynamics and partial measurements[J].Automatica,2014,50(5):1416-1423. [4]YUAN D W,KANHERE S S,HOLLICK M.Instrumenting Wireless Sensor Networks-A survey on the metrics that matter[J].Pervasive and Mobile Computing,2017,37(3):45-62. [5]CATTIVELLI F S,SAYED A H.Modeling Bird Flight Formations Using Diffusion Adaptation[J].IEEE Transactions on Signal Processing,2011,59(5):2038-2051. [6]SHAO T,ZHENG Y R,BENESTY J.An Affine Projection Sign Algorithm Robust Against Impulsive Interferences[J].IEEE Signal Processing Letters,2010,17(4):327-330. [7]NI J G,FENG L.Efficient Implementation of the Affine Projection Sign Algorithm[J].IEEE Signal Processing Letters,2011,19(1):24-26. [8]REN C,WANG Z,ZHAO Z.A New Variable Step-Size Affine Projection Sign Algorithm Based on A Posteriori Estimation Error Analysis[J].Circuits Systems and Signal Processing,2017,36(5):1989-2011. [9]HUANG F Y,ZHANG J S,ZHANG S.Combined-Step-Size Affine Projection Sign Algorithm for Robust Adaptive Filtering in Impulsive Interference Environments[J].IEEE Transactions on Circuits and Systems II:Express Briefs,2016,63(5):493-497. [10]NI J G,MA L S.Distributed Affine Projection Sign Algorithms Against Impulsive Interferences[J].Acta Electronica Sinica,2016,44(7):1555-1560. [11]CHEN B D,LEI X,LIANG J.Steady-State Mean-Square Error Analysis for Adaptive Filtering under the Maximum Correntropy Criterion[J].IEEE Signal Processing Letters,2014,21(7):880-884. [12]WANG S Y,FENG J C,TSE C K.Kernel Affine Projection Sign Algorithms for Combating Impulse Interference[J].IEEE Transactions on Circuits and Systems II:Express Briefs,2013,60(11):811-815. [13]MA W T,CHEN B D,DUAN J D,et al.Diffusion maximum correntropy criterion algorithms for robust distributed estimation[J].Digital Signal Processing,2016,58(2):10-19. [14]WANG W,ZHAO J H,QU H,et al.A correntropy inspired variable step-size sign algorithm against impulsive noises[J].Signal Processing,2017,141(7):168-175. [15]WANG Y,LI Y,BERMUDEZ M,et al.An adaptive combination constrained proportionate normalized maximum correntropy criterion algorithm for sparse channel estimations[J].EURASIP Journal on Advances in Signal Processing,2018,58(1):1-13. [16]MA W T,ZHENG D Q,ZHANG Z Y.Sparse Aware Bias-Compensated Adaptive Filtering Algorithms Using the Maximum Correntropy Criterion for Sparse System Identification with Noisy Input[J].Entropy,2018,20(6):407-421. [17]SHI L,ZHAO H Q.Adaptive Combination of Distributed Incremental Affine Projection Algorithm with Different Projection Orders[J].Circuits Systems & Signal Processing,2018,37(2):1-17. [18]TAKAHASHI N,YAMADA I,SAYED A H.Diffusion Least-Mean Squares with Adaptive Combiners:Formulation and Performance Analysis[J].IEEE Transactions on Signal Processing,2010,58(9):4795-4810. |
[1] | 林云, 黄桢航, 高凡. 扩散式变阶数最大相关熵准则算法 Diffusion Variable Tap-length Maximum Correntropy Criterion Algorithm 计算机科学, 2021, 48(5): 263-269. https://doi.org/10.11896/jsjkx.200300043 |
|