关键词: 非高斯信号处理, alpha谱估计, 最小p范数准则, 分数低阶奇异值分解（SVD）方法, 载波频率检测

Abstract: Least P Norm criterion（LPN） is proposed to improve performance of alpha spectrum estimation based on covariation coefficient matrix and Fractional Lower Order（FLO） covariance matrix；An improved Singularity Value Decomposition（SVD） method，named FLO-SVD，is used to estimate minimal order of AR[SαS]signal model；LPN and IRLS algorithm are identified the parameters of model for estimating alpha spectrum of signal.Simulation results on QPSK signal in impulsive noise show that the proposed method has better performance than methods based on covariation coefficient matrix and FLO covariance matrix and exacter estimation than conventional SVD method in carrier frequency detections.

Key words: nonGaussian signal processing, alpha spectrum estimation, least p norm criterion, fractional lower order Singularity Value Decomposition（SVD） method, carrier frequency detections