摘要
针对小样本条件下的盲信噪比估计误差较大问题,结合信号子空间分解方法,提出一种基于有限样本信息准则(finitesample information criterion,FSIC)的盲信噪比估计算法,并推导出基于FSIC盲信噪比估计算法的最大似然形式。在小样本情况下,FSIC的引入克服了传统信息论方法产生的过拟合和欠拟合问题,降低计算复杂度。在不需要已知信号调制方式、载波频率、波特率等先验知识的前提下,能够在加性高斯白噪声信道(AWGN)和多径信道(Rayleigh)下对常用调制信号进行有效的信噪比估计。在信噪比-25 dB～25 dB范围内,其平均估计误差小于1 dB,表明该算法可有效应用于小样本盲信噪比估计。
Combined with a signal subspace decomposition approach,a new FSIC algorithm with finite sample information criterion was introduced to solve the blind SNR estimates with finite sample in non-cooperative communication,and its likelihood form was also deduced.FSIC can avoid the underfit or overfit produced by the traditional information theory with finite samples,and it can also reduce the complexity.Under the condition of prior-knowledge of received signals unknown,such as modulation types,carrier frequency and baud rate,FSIC can effectively estimate the SNR for communal modulation signals in additional white Gaussian noise（AWGN） channels and multipath channels.The SNR mean estimation error of FSIC is less than 1 dB from-25 dB to 25dB.The results show that FSIC algorithm can be well applied the blind SNR estimate with finite sample in non-cooperative communication.
出处
《科学技术与工程》
北大核心
2012年第2期317-321,共5页
Science Technology and Engineering
基金
国家自然基金重点项目(60634030)
国家自然基金面上项目(61170134
60775012)资助
