用于APSK和QAM信号信噪比估计的改进分段数据拟合算法

为提高幅相键控（APSK）信号和正交调幅（QAM）信号信噪比估计范围和精度,提出了一种改进的信号信噪比估计算法。算法首先计算接收信号平方的均值和绝对值的均值之比,然后根据星座图特征,利用多项式拟合该比值与信噪比的关系。在拟合过程中,对信噪比区间进行分段拟合来提高各段拟合精度,并用蒙特卡洛仿真经验值修正算法的固有偏差,从而得到信噪比的近似无偏估计。仿真结果表明,当信噪比估计区间为-5~20 d B且数据长度合适时,16APSK和32APSK信号信噪比估计偏差均值小于0.5 d B,标准差小于2 d B;该算法对16QAM和32QAM信号信噪比估计的标准差小于传统数据拟合算法。该算法运算复杂度较低...

Full description

Saved in:
Bibliographic Details
Published in电讯技术 Vol. 55; no. 6; pp. 671 - 677
Main Author 许维伟 叶江峰 张伟
Format Journal Article
LanguageChinese
Published 中国工程物理研究院 电子工程研究所,四川 绵阳,621900 2015
Subjects
信噪比估计
分段数据拟合
幅度相位键控
正交幅度调制
盲估计
幅度相位键控
分段数据拟合
subsection data fitting
blind estimation
信噪比估计
APSK
SNR estimation
盲估计
正交幅度调制
QAM
Online AccessGet full text
ISSN1001-893X
DOI10.3969/j.issn.1001-893x.2015.06.015

Cover

More Information
Summary:为提高幅相键控（APSK）信号和正交调幅（QAM）信号信噪比估计范围和精度,提出了一种改进的信号信噪比估计算法。算法首先计算接收信号平方的均值和绝对值的均值之比,然后根据星座图特征,利用多项式拟合该比值与信噪比的关系。在拟合过程中,对信噪比区间进行分段拟合来提高各段拟合精度,并用蒙特卡洛仿真经验值修正算法的固有偏差,从而得到信噪比的近似无偏估计。仿真结果表明,当信噪比估计区间为-5~20 d B且数据长度合适时,16APSK和32APSK信号信噪比估计偏差均值小于0.5 d B,标准差小于2 d B;该算法对16QAM和32QAM信号信噪比估计的标准差小于传统数据拟合算法。该算法运算复杂度较低,便于实时应用和硬件实现,对恒模和非恒模信号均能实现信噪比宽范围精确盲估计。
Bibliography:APSK; QAM ; SNR estimation ; subsection data fitting; blind estimation
To improve the signal-to-noise ratio（SNR） estimation region and precision, an improved SNR estimation algorithm for amplitude phase shift keying（APSK） signals and quadrature amplitude modulation （QAM） signals is proposed. First, the algorithm calculates the ratio of mean value of receiving data＇ s square versus mean value of receiving data＇ s absolute value. Then,the relationship between this ratio and SNR is fitted through polynomial based on the characteristics of constellation map. SNR estimation region is divided to several subsections to improve the precision of data fitting in fitting process and experienced value through Monte Carlo simulation is used to modify the inherent bias of algorithm. Finally, the approxi- mate non-bias estimation of SNR is obtained. When the SNR estimation region is -5 ~ 20 dB and there is enough observation data, simulation results indicate that the SNR estimation mean bias of 16APSK and 32APSK signals is l
ISSN:1001-893X
DOI:10.3969/j.issn.1001-893x.2015.06.015