Research On Spectrum Analysis For Optical Frequency Domain Reflectometry Based On Genetic Algorithm

Optical Frequency Domain Reflectometry(OFDR)is a technique which can achieve accurate detection of optical paths based on frequency-modulated continuous wave(FMCW)radar technique.This paper introduces the development and the working principle and performance index of OFDR,then emphatically discusses the spectrum analysis method of OFDR beat signal,including FFT(Fast Fourier Transform),FFTW(FFT in the West),SFT(Sparse Fourier Transform).The computational steps of the above algorithm are described,and the advantages of the algorithm are pointed out.In dealing with large amounts of data,FFT and FFTW are relatively large amount of computation,there is a bottleneck when improving the frequency resolution of the system.SFT can recover the spectrum sparse signal well,but SFT only restores the frequency components with significant nonzero frequency.For zero frequency components,the output of the SFT is zero-processed,which leads to the loss of some useful measurement information.In this paper,the calculation process of FFT-SCZT(FFT-Segment Chirp-Z Transform)algorithm is discussed in detail which improved the calculation efficiency.There are some shortcomings in FFT-SCZT:First,half-overlap filters are used to improve the detection accuracy and Secondly,the method of locating the peaks by threshold comparison depends on the rationality of the threshold setting,and it is necessary to adjust the threshold according to the different measurement environments.Moreover,when the spectral peak spacing is small,the calculation results overlapping,SCZT will not be able to refine the separation.Through the above discussion,this paper presents a spectrum algorithm based on genetic algorithm for OFDR spectrum analysis named FFT-IGA-ZFFT(FFT-Improved Genetic Algorithm-Zoom FFT).First,calculate a small points FFTW to get a rough spectrum,and then divide the spectrum into groups with a same width,consider each group as a separate individual.Using the improved genetic algorithm(IGA)to search the spectrum peak,the frequency band which needs to be refined is obtained by using the grouping number as the gene,and the sum of the signal values in the group as the fitness.Finally,the band is refined by ZFFT.Considering that there may be multiple effective peaks in the OFDR spectrum,this paper proposes a genetic algorithm with variable parameter iteration to realize the search of multiple effective peaks.Compared with the FFT-SCZT algorithm,the advantages of this algorithm are as follows:no complicated filtering operation is needed,and the calculation amount is reduced;the genetic algorithm is used to search the peak of the spectrum,avoiding the artificially adjusted threshold;using the ZFFT refinement algorithm can solve the problem and reduce the amount of computation.The time complexity of FFT-IGA-ZFFT algorithm and FFT-SCZT algorithm are theoretically deduced and compared.The theoretical analysis shows that FFT-IGA-ZFFT algorithm is more efficient than FFT-SCZT algorithm.In order to verify the effectiveness of the algorithm,this paper constructs an OFDR experimental system to verify the actual running time of the algorithm.The experimental results show that,under the same operating platform,10~7 OFDR samples are processed,the FFT costs 3.13s,the FFT-SCZT costs 1.993s,and the FFT-IGA-ZFFT costs 0.525s to obtain the same precision.It is shown that the algorithm has obvious advantages in processing speed.