Research On The Key Technologies For Reconnaissance Of FH Signals And Prediction Of FH Sequence

Frequency hopping(FH)communication is a typical mode of spread spectrum communication.It has been widly applied not only in military communication,but also in civilian mobile communication.Under the control of FH sequence,the carrier frequency of the signal hops in the wide band pseudo-randomly,by which the spectrum is expanding.FH signal has the characteristics of anti fading,anti-jamming,anti interception,easy networking and strong multiple access.Moreover,with the increase of FH rate and FH bandwidth,the FH communication is facing with more severe challenges.At present,the research of FH signal reconnaissance algorithmes mainly focus on the accurate estimation of FH signal parameters,the real-time separation of multiple FH signals and the compression sampling of broadband FH signals.At the same time,as an auxiliary mothod of FH signal reconnaissance,the prediction of FH sequence has great significance for improving the real-time performance of FH signal reconnaissance and the efficient reconnaissance and tracking jamming.In this paper,the four problems of FH signal reconnaissance and FH sequence prediction are studied: the semiblind detection and tracking of FH signal,the detection and frequency estimation of FH signal based on modulated wideband converter(MWC),tracking and separation of FH signal based on dynamic programming MWC and prediction and separation of chaotic frequency hopping sequences based on radial basis function(RBF)neural network.The main research contents are detailed as follows:In Chapter 2,based on the theory of FH signal communication,a specific FH signal frame structure is proposed and corresponding channel coding and modulation methods are selected as the signal model in this chapter and other chapters.Then,according to the signal model,a semiblind detection algorithm and tracking strategy of FH signal based on the time-frequency analysis method with variable window length are proposed.Firstly,the hopping of FH signal is detected by linear time-frequency analysis of short-time window.Then,the carrier frequency of current FH signal is accurately estimated by combined time-frequency analysis method of long-time window.Finally,the synchronous prefix is demodulated for tracking and verifing and a complete semiblind detection and tracking system of FH signal is constructed.The simulation results show that the algorithm can detect and estimate the carrier frequency of the current FH signal quickly.In Chapter 3,the problem of FH signal detection and frequency estimation based on compressive sensing is studied.Based on the sparsity and short-time stationary characteristics of FH signal in time-frequency domain,a detection algorithm of FH signal based on channelized MWC and a method for carrier frequency estimation without reconstruction are proposed.Firstly,periodic waveform of each channel is reconstructed in MWC and the channelized MWC structure is obtained.Then,the real-time detection of FH signal is realized in frequency domain based on characteristics of recognition.Finally,the carrier frequency of FH signal is estimated by the compressive baseband data without reconstructing the signal.The algorithm effectively reduces the sampling rate and the complexity of operation.Simulation results verify the effectiveness of the algorithm.In Chapter 4,the problem of tracking and separation for FH signals based on compressive sensing is studied.A frequency hopping signal tracking and separation algorithm based on dynamic programming MWC is proposed.Firstly,the multi frequency function of feedback control is added to the MWC system,and the dynamic programming MWC structure is obtained.Then the energy of MWC sub-channel is detected in time domain to track FH signal in real-time.Finally,the time of hopping information and the power difference of different signal sources are used to separate the FH signals.The method assumes that the power of the signal from the same FH source is relatively stable to the receiver.Simulation results verify that the methods can track and separate the multiple FH signals effectively.In Chapter 5,the prediction and separation of chaotic FH sequences are studied from the perspective of FH sequence characteristics.Based on the random orthogonality and global phase diagram of chaotic frequency hopping sequences,the radial basis function(RBF)neural network is trained to predict and separate the mixed frequency hopping sequences.Firstly,the characteristics of chaotic FH sequences are analyzed,and the input and output for the training of neural network are determined by K-means clustering algorithm.Then,the embedding dimension of chaotic sequence is calculated by the orthogonal least square(OLS),and the neural network is constructed.Finally,the prediction and separation of mixed sequences are realized by segment matching methods.Simulation results show that the algorithm can predict the chaotic frequency hopping sequences in synchronous networks and separate the mixed frequency hopping sequences effectively.