摘要
本文在最大熵谱分析法甚础上,导出了一种自相关检测器。该检测器结构简单,并且有Robust性和自回归(AR)检测器的大多数优点。通过解析计算和计算机模拟,研究了该检测器在白噪声中的信号检测能力,并与传统的理想匹配滤波器组检测性能进行了比较。研究结果表明:在高斯白噪声中,当样本数较少(N<64)时,与理想匹配滤波组检测器相比,自相关检测器的小样本性能的信噪比损失在3db之内。在非高斯白噪声中,信噪比改善比白高斯噪声情况下更为显著。并且其检测性能随噪声参数的变化要比理想匹配滤波器组检测器小。由于结构简单,该检测器所需的计算量度存贮器数目都较匹配滤波器组检测器少。因而硬件实现较为简便。
Based on Maximum Entropy Spectral Analysis, a kind of autocorrelation detector is deduced in this paper. In addition to the simple structure, this detector has Robust property and most of the advantages of AR detector. tly theoretic analysis and Monte-Carlo simulation, this paper examines the de Bction performance of the detector in white noise, and compares itsperformancwith that of the ideal Matched Filter Bank (i e FFT)detector. The results show: in white Gaussian noise, under small-sample Condition (N<64), compared with the ideal FFT detector, the signal-noise-ratio(SNS) loss of the detection performance, of the detector is within 3db. In non-Gaussian white noise, the SNR improvement is more remarkable. And its detection perfhormance is less sensitive to the noise parameters than that of the FFT detector.Moreover, the substantial savings in Computation and storage make the ardware realization much more simple and convenient.
出处
《信号处理》
CSCD
北大核心
1990年第2期65-73,81,共10页
Journal of Signal Processing
