Research On Synthesis Method Of Multi-subband Radar Signals

With the continuous progress of radar systems,in military and civilian fields,radar resolution plays an important role in radar detection and recognition.According to radar range resolution theory,range resolution is limited by signal bandwidth.In order to obtain better resolution,signal usually requires a large bandwidth.According to Nyquist sampling theorem,the increase of signal bandwidth will cause sampling frequency to increase at least double.In order to obtain ultra-wideband radar signal designed with high precision range resolution,it will bring huge pressure on transmitter and receiver and other equipment,which consumes a lot of hardware costs.This paper firstly introduces a multi-subband synthesis algorithm that can effectively improve the resolution.This method constructs a parametric model of the signal,and calculates the model parameters to predict blank subbands,thereby synthesizing subbands into a full-band signal.At present,many researches believe that multi-subband synthesis is to achieve the improvement of range resolution through the expansion of the bandwidth,and this paper proposes a new understanding of multi-subband synthesis resolution,the Inverse Discrete Fourier Transform(IDFT)of the baseband signal.It can represent range envelope of the target,and its resolution is affected by the frequency resolution.The essence of multi-subband synthesis is the expansion of the baseband signal time width to improve frequency resolution.Then,a noise suppression algorithm for pole discriminant inaccuracy of all-pole model in low Signal-Noise Ratio(SNR)environment is introduced.All-pole model adopted by traditional subband synthesis method uses Root-MUSIC(Multiple Signal Classification)to calculate pole values,in which poles are selected as pole values closest to unit element,but when attenuation terms such as noise are introduced,the required poles values will deviate from the unit circle,and the interference poles that deviate from unit circle are closer to unit circle,resulting in errors in pole selection and error in prediction model estimation.In order to deal with this problem,this paper proposes a noise suppression method combined with subband model.Through weighting and placement of each component of main singular value of signal Hankel matrix in the process of pole calculation,multi-frequency noise in a low-noise environment is realized subband synthesis.Although methods such as noise suppression can make up for the insufficiency of the allpole model in dealing with noise,there are still many defects.For example,there are too many complex links in traditional algorithms,and a single algorithm can test a single sample,which is not conducive to long-term detection of targets.Furthermore,approximating a linear model to a nonlinear signal creates inevitable errors.Finally,in order to solve the above problems,this paper proposes to use Deep Neural Network(DNN)to realize multi-subband synthesis.By using the distance envelope of the multi-subband signal as the input of the DNN,and using the distance envelope of the full-band signal as the label,the process from the multi-subband distance envelope to the full-band distance envelope is realized.Therefore,the multi-subband synthesis algorithm for synthesizing multiple pulse signals into a broadband signal has broad application prospects.