Research On Fast Parameter Estimation Technology Of Linear Frequency Modulated Signal

As a typical non-stationary signal, linear frequency modulated(LFM) signal is widelyused in various practical applications such as sonar, radar and underwater acousticcommunication, therefore, the parameter estimation of LFM signal was a hot research issueall the times. There are many estimation methods, but some of them have a contradictionbetween high accuracy and low computation, so it’s hard to achieve real-time processing inthe existing hardware conditions. So it’s significant to find fast estimation algorithms on thepremise of accuracy. The paper studies the problem of large computation for dechirp methodand Fractional Fourier Transform(FRFT). The main contents are as follows:Firstly, three methods of LFM signal parameter estimation—delay autocorrelation,dechirp and FRFT are studied. This paper has been concerned with deducing the Ozaktas’ salgorithm and the estimation formulas for LFM signal parameter estimation based on FRFT.Secondly, according to the delay autocorrelation method with low accuracy andcomputation, and the dechirp method with high accuracy and computation, a modified dechirpmethod is proposed. The simulation shows that the computation of this modified method ismuch lower than the dechirp method.Thirdly, since Particle Swarm Optimization(PSO) algorithm gets into local optimalsolution easily when processing function with many local extremums, we propose animproved PSO algorithm whose initialization and iterative formula are changed. Since themethod by fixed step FRFT has a contradiction between high accuracy and low computation,we propose a new method which combines the improved PSO and FRFT. The simulationshows that the proposed new method has a much better performance than the method by fixedstep FRFT when SNR(Signal Noise Ratio) is higher than-12.5 d B, and it eases thecontradiction between high accuracy and low computation effectively.