Study On Target Localization Algorithms Of Bistatic MIMO Radar Based On Time Frequency Analysis

Multiple input multiple output (MIMO) radar has been paid a lot of attention bymany researchers since it came out. It makes up for the deficiency of the conventionalradar system, and obtains good gains of detection and estimation by making full use ofdiversity or coherent processing in both of the transmit array and receive array. Targetlocalization has been widely studied in bistatic MIMO radar. The ways to realizetarget localization are to utilize joint DOD and DOA estimation and triangle location.MIMO radar can improve the resolution of DOD and DOA estimation by forming alarge virtual array aperture. However, it usually assumes that the echo signal model ofthe target is stationary signal for MIMO radar. However, in the actual cases, radarsignals are often nonstationary with Doppler frequencies that may change with time,which called nonstationary signal. Time frequency analysis is a powerful tool fornonstationary signal processing, and the effective combination with the array signalprocessing, which leads the time-varying nature of the nonstationary signal to beshown accurately when time-delay and frequency-offset are used to describe therelation between signals. It can also make the analysis of the nonstationary signalcorrect to the specific frequency during a particular time, a good time frequencyfeatures can be obtained to dramatically improve the performance of the space arraysignal processing.Based on time frequency analysis, the paper has a deeply research on the methodof target localization in bistatic MIMO radar. The main innovations of the research areas follows:Regard the nonstationary echo signal of the target as a LFM signal, a bistaticMIMO radar nonstationary signal model based on uniform linear array is described.An algorithm called ESPRIT-RootMUSIC method based on ambiguity function isproposed, by constructing a spatial time frequency distribution matrix, usingESPRIT-RootMUSIC method to realize the joint DOD and DOA estimation. Themethod does not need multiple-dimensional spectrum peak searching and theparameters can be paired automatically. The simulation results demonstrate theeffectiveness of the algorithm. When compared with the method based on pseudo Wigner-Ville distribution, the advantages of extracting the auto term and removing thenosie are more effectively which based on ambiguity function in this paper. Inaddition, the accuracy of localization is better than the algorithm based on pseudoWigner-Ville distribution under the condition of fewer snapshots.Most of the parameter estimation algorithms have been proposed by now, usuallydealt with the received data directly in the complex domain, which greatly increase theamount of calculation of the algorithm. We put forward a Unitary-ESPRIT methodbased on ambiguity function (named as AF-UESPRIT in this thesis), by reconstructingthe received data to increase the information of the data, and then, exploiting atransformation matrix to change the complex domain of the original array into the realones. The computational complexity of the algorithm can be reduced when analysisand process the real number of time-frequency matrix, and the parameters of DODand DOA can be paired automatically. The effectiveness of the proposed algorithmcan be verified by the following simulation.The estimation performance of traditional methods can be reduced when thetargets are spaced closely in spatial domain. In order to realize the precise localizationof the targets, the time frequency analysis method is applied to the polarimetricbistatic MIMO radar. A uniform polarized linear receive array bistatic MIMO radarmodel for nonstationary signal is established. An algorithm for the polarimetricbistatic MIMO radar based on ambiguity function is proposed. Utilizing thepolarization diversity, the resolution of the DOD and DOA estimation can beimproved obviously. The simulation results verify the effectiveness of the algorithm,especially, the algorithm can distinguish the targets which have the similarDOD/DOAparameters to realize the target localization with high resolution.