Study On Anti-interference Waveform Design For Cognitive Radars

With the increasing complexity of the electromagnetic environment,the antiinterference technology of traditional radar performs poorly due to the lack of the ability to learn and utilize the environmental knowledge.Cognitive radar changes the fixed working mode of traditional radar.Based on the closed loop of information feedback from the receiving end to the transmitting end,cognitive radar can utilize the interference environment knowledge to optimize the transmitted waveform,thereby improving the cognitive and adaptive capability.This is expected to break through the anti-interference bottleneck of traditional radar.The research in this paper aims at transforming passive,static,and artificial anti-interference methods into active,dynamic,and intelligent anti-interference methods.Specifically,this paper carries out research on LPI waveform design,KB artiinterference waveform design,and dynamic learning-based anti-interference waveform design.The details are as follows.1.To counter advanced interception technologies of active jammers,a new waveform of LFM-BPSK with a varying chirp rate(denoted as VCR-LFM-BPSK)is designed.In this design,based on the interception frame of the jammer,different chirp rates are introduced in different sub-pulses to prevent the signal from being intercepted.Then,the chirp rates of the VCR-LFM-BPSK signal are optimized by minimizing the peak level of the dechirped signal to counter the dechirp technology of the jammer.Moreover,the binary phases of the VCRLFM-BPSK signal are designed via a multiobjective Pareto optimization to improve the autocorrelation capability and lower the probability of interception in the frequency domain.Numerical examples verify that the designed hybrid signal outperforms conventional radar signals in countering interception technologies and possesses better target detection performance.2.To counter clutter and active interference in main and side lobes at the same time,a KB space-time anti-interference waveform design method based on digital array radar is proposed.The design problem is modeled by maximizing the output signal-to-interferenceplus-noise ratio(SINR)with the constraints of spectrum compatibility and constant modulus.To solve the resulting complicated optimization problem,the hierarchical approach is employed to convert the original problem into two subproblems,i.e.,submatrix design for sidelobe interference suppression and diagonal matrix design for countering mainlobe interference.The first subproblem is solved by a novel anti-sidelobe-interference algorithm named hybrid beampattern synthesizing(HBS)which can achieve lower nulling levels than the well-known beampattern synthesizing methods.The second subproblem is solved by two steps composed of energy spectrum density(ESD)design and signal design via the proposed hybrid ESD matching(HEM)algorithm which can balance matching loss and spectrum compatibility more flexibly than the existing ESD matching methods.Numerical examples show that the proposed design has better anti-interference performance than existing algorithms.3.To avoid the degradation of waveform autocorrelation ability caused by antiinterference performance optimization,an anti-interference waveform optimization model under autocorrelation constraint is proposed.In contrast to several counterparts available in the open literature,the waveform design described herein is formulated by optimizing the output SINR under multiple constraints of spectrum compatibility,transmit energy,a constant modulus,and an autocorrelation function.To resolve the resulting intractable optimization problem,a time-frequency-time solution procedure is developed.First,the waveform optimization model in the time domain is transformed into an ESD optimization model in the frequency domain.Then,the double-search-based technique is proposed to obtain the ESD solution,thus the waveform in the time domain can be synthesized by ESD matching methods.The waveforms designed by conventional ESD matching algorithms have unregular ESD spikes,which will decrease the autocorrelation function and LPI performance of the transmit waveform.To solve this problem,two ESD matching methods are proposed,including the adaptive alternating projection-based phase-modulated waveform synthesis(AAPWS)method and the simulated annealing(SA)-based linear frequency modulated waveform synthesis(LFMWS)method.In the analysis stage,the effectiveness of the proposed ESD solving and matching methods is explained.Besides,the achieved ESD shape,autocorrelation function,and output SINR are presented to verify the performance of the proposed method.4.Aiming at the problem of anti-interference performance degradation of cognitive radar caused by inaccurate interference knowledge in a dynamic active jamming environment,an anti-interference waveform design based on interference parameter prediction is proposed.Specifically,an OS-ELM-based interference frequency prediction(OS-ELM-FP)model and an OS-ELM-based interference angle prediction(OS-ELM-AP)model are proposed to learn the interference activities in the frequency and spatial domains,respectively.Benefiting from the single-hidden-layer network structure and recursiveformula-based output weight updating,OS-ELM-FP and OS-ELM-AP possess the ability to predict the interference state and update the prediction model parameters based on the dynamically updated interference dataset with much higher computational efficiency.Simulation and measured interference data verify that the proposed method can achieve lower prediction errors with a low computational complexity compared with the state-ofthe-art methods.Furthermore,it is shown in cognitive anti-interference experiments that the proposed method significantly improves the anti-interference performance of cognitive radars by predicting active interference knowledge.5.Aiming at the problem of insufficient cognitive learning ability of traditional antiinterference methods in dynamic active interference environment,a reinforcement learning based cognitive frequency design method for compressed sensing based frequency agile radar(CS-based FAR)is proposed.First,we formulate the frequency design as a model-free partially observable Markov decision process(POMDP)to cope with the non-cooperation of the active interference environment.Then,a recognizer-based belief state computing method is proposed to relieve the storage and computation burdens in solving the POMDP.This method is independent of the environmental model knowledge and robust to the sensing scenario.Finally,the double deep Q network(DDQN)-based method using the exploration strategy integrating the CS-based recovery metric into the ε-greedy strategy(DDQN-CSRε-greedy)is proposed to solve the POMDP.The experimental results demonstrate that the designed DDQN-CSR-ε-greedy algorithm can successfully learn a variety of dynamic interference schemes and can obtain better target measurement performance while avoiding interference.