Research On Frequency-domain Adaptive Line Enhancement Based On Matched Filter

In electronic information systems of sonar,radar,and communications,the matched filter(MF),which is the most basic theory and the most classic detector in the field of signal detection,is always the most commonly used among a variety of signal detection methods.It is proved that the MF is the optimal linear filter that obtains the maximum output signal to noise ratio(SNR)to detect the known signals in white ambient noise(the so-called ideal conditions).In order to challenge this optimality,a frequency-domain adaptive line enhancement method based on matched filter(FDAMF)is proposed in this thesis to find a better detector than the MF in the field of adaptive filters under ideal conditions.It will provide more reliable and efficient detection techniques to engineering applications.The basic idea of implementing the FDAMF is to prove that the frequency-domain output of the MF contains a periodic component by analyzing the output spectrum characteristics of the MF and transform the output spectrum of the MF into a frequency-domain line detection problem.Based on the theoretical analysis,we can use adaptive line enhancement methods to improve the performance of the MF.Based on this idea,the thesis first studies the line enhancement techniques and proposes a time reversal convolution and interference suppression(TRCIS)method.We prove that the physical meaning of the time reversal convolution(TRC)is to convert the correlation matrix into a sequence which contains all elements of the correlation matrix.Then we derive the principle of the TRCIS for detecting CW signal in broadband noise and suppressing multipath interference.From the simulation experiments,conclusions can be drawn such as the signal-to-noise ratio(SNR)gain of TRCIS increases linearly with the input SNR;the detection threshold is about 2 dB lower than that of the conventional line detection methods;the receiver operator characteristic(ROC)curve is significantly better than that of the conventional line detection methods.Then,by using TRCIS as a preprocessing technique of adaptive line enhancer(ALE),we propose an improved ALE,that is TRCIS-ALE,and present the implementation principle.From the simulation comparison with conventional line detector,ALE and coherent accumulation based adaptive line enhancer(CA-ALE),we can obtain the conclusion that the SNR gain,detection threshold and ROC curve of TRCIS-ALE are all significantly better than the other three methods.Combining the above ALE with MF,a frequency-domain adaptive line enhancement method based on matched filter(FDAMF)is proposed.For lower input SNR,a TRCIS-ALE based FDAMF is proposed combining with the TRCIS-ALE and their implementation principles are presented.It can be proved that the FDAMFs have better detection performance than the conventional MF from the simulation results.For instance,the SNR gain of FDAMF is much larger than that of MF,the detection threshold of FDAMF is almost 2 dB lower than that of MF,and the detection probability of FDAMF is approximately 20% higher than that of MF with the same input SNR and false alarm probability.According to the characteristics of sonar receivers,several dual-input FDAMFs(DIFDAMF)are proposed,such as split-array dual-input FDAMF(SA-DIFDAMF),positive-negative-frequency dual-input FDAMF(PN-DIFDAMF)and real-imaginary-part dual-input FADMF(RI-DIFDAMF).From the simulation experiments,it is shown that SA-DIFDAMF has better detection performance than the other two methods within the range of detection threshold.The adaptability of FDAMF is studied for some practical problems in the active sonar systems.For narrowband signal,we process it in baseband by using Hilbert transform to reduce the computational complexity and the implementing difficulty of prefilters,and use sample-cycling adaptive technique to widen the bandwidth of the signal artificially,and then validate the performance of the measures through simulation experiments.For different sonar waveforms,FDAMF has better detection effects than the conventional MF,which indicates that FDAMF have wide applicability for different waveform signals.For colored noise background,we use pre-whitening method to achieve pre-whitening FDAMF,and validate the detection performance of pre-whitening FDAMF is better than FDAMF and pre-whitening MF.For coherent multipath channel,we propose a model-based FDAMF by estimating the channel impulse response to modify the reference signal,and validate the detection performance of model-based FDAMF is better than FDAMF and model-based MF.In order to further verify the actual effect of the proposed method,the actual experiment data are processed and analyzed.The results show that: TRCIS-ALE can effectively suppress background interference and preserve the original line spectrum information to the greatest extent in line detection of ship radiation noise.The actual echo detection effect of the FDAMF of broadband LFM signal is much better than that of the conventional MF,and FDAMF effectively suppresses the interference noise and improves the processing gain of MF.The FDAMF for narrowband signals significantly improved detection performance of the MF of narrowband HFM signal,which demonstrates the effectiveness of FDAMF for narrowband signal echo detection.The results of these experiments provide the experimental basis for the actual engineering application of FDAMF.