传统的最小均方误差(Least Mean Square,LMS)算法具有计算量低、易于实现的优点,广泛应用于各种场景,但其平均收敛速度和稳态误差相互制约。基于箕舌线函数(Tongue Line)的变步长LMS(Tongue Line Least Mean Square,TLLMS)算法引入箕舌...传统的最小均方误差(Least Mean Square,LMS)算法具有计算量低、易于实现的优点,广泛应用于各种场景,但其平均收敛速度和稳态误差相互制约。基于箕舌线函数(Tongue Line)的变步长LMS(Tongue Line Least Mean Square,TLLMS)算法引入箕舌线函数,改进了步长更新公式,从而提高了收敛速度并降低了稳态误差。然而,该算法的步长易受噪声干扰,导致滤波器权值在最优权值附近波动。为此,文章提出了改进箕舌线变步长LMS(Improved Tongue Line Least Mean Square,ITLLMS)算法。该算法进一步优化了步长更新公式,消除了输入端不相关噪声的干扰,以最小化收敛时的稳态误差,并加入上一步步长的影响以增强系统的跟踪能力。将ITLLMS算法应用于自适应噪声抵消系统的实验结果表明,与基于相关特性(Correlation Characteristic)的箕舌线变步长LMS(Correlation Characteristic Tongue Line Least Mean Square,CCTLLMS)算法相比,文章提出的ITLLMS算法具有更快的收敛速度、更低的稳态误差以及更低的计算量。展开更多
It is very important for high temperature superconducting electronic devices to increase the grain size of YBCO epitaxial films because it can effectively reduce the defects and improve the probability of successful p...It is very important for high temperature superconducting electronic devices to increase the grain size of YBCO epitaxial films because it can effectively reduce the defects and improve the probability of successful preparation of Josephson junction.In this study,YBa_(2)Cu_(3)O_(7-δ)(YBCO)films with grain size in excess of 1.5μm were successfully prepared by the directly heating SrTiO_(3)substrates coated by Si C on their back.Interestingly,the grain size of YBCO film is enhanced greatly by this directly heating method,and the critical temperature TCand critical current density JCof YBCO films are as high as 91.5 K and 3.5 MA/cm^(2),respectively.Compared with the traditional indirect heating method,which involves applying silver paste and then using a heat soaking block(e.g.Inconel 600),this direct heating method effectively enhances the grain size of YBCO film and the possibility of successful preparation of Josephson junction.展开更多
A Matrix Inversion Normalized Least Mean Square (MI-NLMS) adaptive beamforming algorithm was developed for smart antenna application. The MI-NLMS which combined the individual good aspects of Sample Matrix Inversion (...A Matrix Inversion Normalized Least Mean Square (MI-NLMS) adaptive beamforming algorithm was developed for smart antenna application. The MI-NLMS which combined the individual good aspects of Sample Matrix Inversion (SMI) and the Normalized Least Mean Square (NLMS) algorithms is described. Simulation results showed that the less complexity MI-NLMS yields 15 dB improvements in interference suppression and 5 dB gain enhancement over LMS algorithm, converges from the initial iteration and achieves 24% BER improvements at cochannel interference equal to 5. For the case of 4-element uniform linear array antenna, MI-NLMS achieved 76% BER reduction over LMS algorithm.展开更多
论文研究了自适应最小均方误差(Least Mean Squares,LMS)滤波算法的步长选取问题。在分析现有算法的基础上,通过构造步长与误差信号之间的非线性函数,提出一种新的变步长LMS算法。新算法采用误差信号的自相关估计值控制步长,而不是直接...论文研究了自适应最小均方误差(Least Mean Squares,LMS)滤波算法的步长选取问题。在分析现有算法的基础上,通过构造步长与误差信号之间的非线性函数,提出一种新的变步长LMS算法。新算法采用误差信号的自相关估计值控制步长,而不是直接利用瞬时误差控制步长,避免了噪声干扰,降低了稳态失调,可工作于低信噪比环境。同时新算法步长控制无记忆效应,提高了收敛速度。仿真表明,新算法的稳态失调和收敛速度均优于现有变步长LMS算法。展开更多
文摘传统的最小均方误差(Least Mean Square,LMS)算法具有计算量低、易于实现的优点,广泛应用于各种场景,但其平均收敛速度和稳态误差相互制约。基于箕舌线函数(Tongue Line)的变步长LMS(Tongue Line Least Mean Square,TLLMS)算法引入箕舌线函数,改进了步长更新公式,从而提高了收敛速度并降低了稳态误差。然而,该算法的步长易受噪声干扰,导致滤波器权值在最优权值附近波动。为此,文章提出了改进箕舌线变步长LMS(Improved Tongue Line Least Mean Square,ITLLMS)算法。该算法进一步优化了步长更新公式,消除了输入端不相关噪声的干扰,以最小化收敛时的稳态误差,并加入上一步步长的影响以增强系统的跟踪能力。将ITLLMS算法应用于自适应噪声抵消系统的实验结果表明,与基于相关特性(Correlation Characteristic)的箕舌线变步长LMS(Correlation Characteristic Tongue Line Least Mean Square,CCTLLMS)算法相比,文章提出的ITLLMS算法具有更快的收敛速度、更低的稳态误差以及更低的计算量。
基金Project supported by the National Key Research and Development Program of China(Grant No.2023YFF0720500)the State Key Program of the National Natural Science Foundation of China(Grant No.U22A2019)the National Key Project(Grant No.22-05-CXZX-04-03-15)。
文摘It is very important for high temperature superconducting electronic devices to increase the grain size of YBCO epitaxial films because it can effectively reduce the defects and improve the probability of successful preparation of Josephson junction.In this study,YBa_(2)Cu_(3)O_(7-δ)(YBCO)films with grain size in excess of 1.5μm were successfully prepared by the directly heating SrTiO_(3)substrates coated by Si C on their back.Interestingly,the grain size of YBCO film is enhanced greatly by this directly heating method,and the critical temperature TCand critical current density JCof YBCO films are as high as 91.5 K and 3.5 MA/cm^(2),respectively.Compared with the traditional indirect heating method,which involves applying silver paste and then using a heat soaking block(e.g.Inconel 600),this direct heating method effectively enhances the grain size of YBCO film and the possibility of successful preparation of Josephson junction.
基金Project supported by the IRPA Secretariat, Ministry of Science,Technology and Environment of Malaysia (No. 04-02-02-0029) andthe Zamalah Scheme
文摘A Matrix Inversion Normalized Least Mean Square (MI-NLMS) adaptive beamforming algorithm was developed for smart antenna application. The MI-NLMS which combined the individual good aspects of Sample Matrix Inversion (SMI) and the Normalized Least Mean Square (NLMS) algorithms is described. Simulation results showed that the less complexity MI-NLMS yields 15 dB improvements in interference suppression and 5 dB gain enhancement over LMS algorithm, converges from the initial iteration and achieves 24% BER improvements at cochannel interference equal to 5. For the case of 4-element uniform linear array antenna, MI-NLMS achieved 76% BER reduction over LMS algorithm.
文摘论文研究了自适应最小均方误差(Least Mean Squares,LMS)滤波算法的步长选取问题。在分析现有算法的基础上,通过构造步长与误差信号之间的非线性函数,提出一种新的变步长LMS算法。新算法采用误差信号的自相关估计值控制步长,而不是直接利用瞬时误差控制步长,避免了噪声干扰,降低了稳态失调,可工作于低信噪比环境。同时新算法步长控制无记忆效应,提高了收敛速度。仿真表明,新算法的稳态失调和收敛速度均优于现有变步长LMS算法。
