| Meterwave radar,also called VHF radar has the advantages of long distance detection,anti stealth and anti interference etc.But at the same time,because it has a broad beam,on one hand,the direct wave and reflected wave are indistinguishable in the same beam,and on the other hand,the existence of "ground beam" causes the beam split,all of this make VHF radar measure target elevation of an deviation at low elevation areas.The existing subspace super-resolution algorithm can solve the above problems,but the computation is large,and it can not meet the real-time requirements.Some subspace super-resolution algorithms also need to match the suitable terrain to measure the elevation of targets accurately.Aiming at solving the above problems,the Digital Beam Forming algorithm,the Synthetic Steering Vector Maximum Likelihood algorithm and the Modified Generalized Constrained MUSIC algorithm are introduced on this thesis,then the performance analysis and realizes it through GPU are introduced too.Finally,the terrain correction algorithm is studied and the simulation analysis is given.The specific work of the thesis is as follows:Firstly,the GPU high-performance computing and Compute Unified Device Architecture programming models are introduced.The CUDA programming model is GPU+CPU heterogeneous computing model,which GPU is responsible for the operation of parallel dense data and CPU is responsible for data initialization and process control.The VHF radar angle measuring involves many matrix operation,so how to call the CUDA library function is introduced.Finally,the methods of CUDA performance optimization are introduced,including data transmission bandwidth optimization,memory optimization and instruction flow optimization between CPU and GPU etc.Secondly,the commonly used signal reception models are introduced,which take the multipath signal and the earth curvature into account,it's also improve the accuracy of angle measurement.On this basis,the formation of the Digital Beam Forming algorithm,the Synthetic Steering Vector Maximum Likelihood algorithm and the Modified Generalized Constrained MUSIC algorithm is introduced,the Synthetic Steering Vector Maximum Likelihood algorithm and the Modified Generalized Constrained MUSIC algorithm are analyzed,the dimension of two algorithms is reduced according to the relationship of the incident angle and the reflection angle,which greatly reduces the amount of computation.Then the three algorithms are implemented on GPU and CPU respectively.The detailed flow of the algorithm on the GPU and the performance comparison analysis on the two platforms are given.Considering the Digital Beam Forming algorithm has simple principle and fast calculation speed,but the Synthetic Steering Vector Maximum Likelihood algorithm is more complex and slow computing speed.So the two algorithms are used to measure elevation angle at the time of realization,which means we first use the Digital Beam Forming algorithm to measure the range of the target's elevation.Then we use the Synthetic Steering Vector Maximum Likelihood algorithm to search for the target's real elevation angle in this range,so as to improve the computing speed further.Finally,for the Synthetic Steering Vector Maximum Likelihood algorithm needs to be matched with the appropriate terrain in order to estimate the target elevation angle accurately,a terrain correction method based on ADS-B and DEM is proposed on this thesis,that need to convert the ADS-B data format and store it with the DEM data firstly.In the same array,the ADS-B data is then matched with the terminal data,and then the target height is calculated using a Synthetic Steering Vector Maximum Likelihood algorithm,and then the height is compared with the corresponding DEM elevation,and finally the relationship between the elevation error and DEM is corrected.This method can complement the ADS-B method and perform information correction without civil aviation and high maneuvering.The effect is better than no modification. |
PDF Full Text Request