Parameter Estimation of FH Signals Based onαStable Noise Sparsity and Optimal Match

Yan JIN,Yanfeng WU,Hongbing JI

DOI: https://doi.org/10.11999/JEIT161397

2017-01-01

Abstract:Currently, FH signal parameter estimation methods based on compressed sensing are mostly under the assumption of Gaussian noise background. In non-Gaussianα stable distribution noise conditions, the algorithms based on Gaussian noise model suffer undesirable performance degradation. In this paper, it is analyzed and concluded that the spike pulses of theα stable noise approximately meet sparse conditions. By using the differences of the characteristics in the time domain, the FH signal and the noise can be easily separated, and the goal of suppressing noise can be achieved. Under the framework of compressed sensing, the three-parameter dictionary is constructed based on the characteristics of FH signals, then the Optimal Match (OM) for adaptive FH signal decomposition is used to obtain the matching atoms and the FH signal parameters are estimated based on the information contained by these time frequency atoms. Simulation results show that compared with the conventional FH signal parameter estimation methods, the proposed Sparsity-OM (SOM), which uses noise sparsity to suppress the noise and then adopts the OM algorithm, improves the estimation accuracy of FH signal parameters and it is more robust to theα stable distribution noise.