基于GP分布拟合检验的LFM/BPSK处理结果可信性评估

Credibility Evaluation for Processing Results of LFM/BPSK Signals Based on Goodness-of-Fit Test of Generalized Pareto Distribution

研究了线性调频/二相编码(LFM/BPSK)混合调制信号处理结果的可信性评估问题,提出了一种基于相关谱超阈值分布拟合优度检验的评估算法.根据某一次识别处理结果及其对应的调制信号模型,建立参考信号,并计算其与原始观测信号的相关谱,通过对相关谱极大值概率分布特性的分析,将LFM/BPSK信号处理结果的可信性评估转化为对相关谱超阈值序列概率分布的拟合优度检验问题.文中利用极大值分布吸引场理论,证明了零假设下相关谱超阈值的超出量近似服从广义Pareto分布.计算机仿真结果验证了本算法的有效性.

A credibility test method,based on the goodness-of-fit test of the exceedance distribution of the cross-correlation spectrum,is proposed to evaluate blind processing results of LFM/BPSK hybrid modulation signals.A reference signal is constructed depending on the identified modulation format and the corresponding estimated parameters,and the spectrum of the cross-correlation between the observed signal and the constructed reference signal is calculated.By using the property of the maximum distribution of the cross-correlation spectrum,the credibility test is transformed to the goodness-of-fit test of the exceedance distribution to the cross-correlation spectrum.It is proved that the exceedance distribution under null hypothesis is approximated to a generalized Pareto distribution according to the maximum distribution attraction theory.The effectiveness of the algorithm is demonstrated in extensive simulations.