The Design And Implementation Of Hardware-in-the-Loop Simulation System For Forward-looking Imaging Based On GPU

The traditional Synthetic Aperture Radar(SAR) and Inverse Synthetic Aperture Radar(ISAR) are parts of coherent imaging radar system with high resolution, which can take advantage of Doppler information from the relative motion between target and radar to get the imaging result. And both of them can work in all weather conditions, day and night, which makes them widely applied in many fields. However, when the situation in front of radar carrier is needed to detect, it will give rise to the blind zone problem in forward-looking mode.For the purpose of overcoming this problem mentioned above, a large number of schemes are proposed. In recent years, with the development of quantum optics, especially the advent of computational ghost imaging technology, a new kind of forward-looking imaging algorithm is presented, which is called the forward-looking microwave correlated imaging algorithm based on the quantum theory. Although it has excellent resolution, with the disadvantage of huge computation and high-level hardware requirements, traditional signal processing platform can not implement and test this algorithm in a short period of time. With the rapid development of electronic technology, Graphics Processing Units(GPU) have become more and more powerful, which makes its application expand from graphical display to general signal processing field. In view of this, this thesis makes full use of GPU’s parallel capability to optimize the imaging algorithm, which greatly shortens the running time.This thesis mainly focuses on the achievement of forward-looking microwave correlated imaging algorithm on GPU and verification of hardware-in-the-loop simulation system. Firstly, the basic concept of quantum correlated imaging algorithm and compressive sensing theory are introduced. Then the discussion about sparse signal reconstruction algorithm is given in particular, and the signal model and processing flow of forward-looking imaging algorithm are presented. According to the scheme performance requirements, a minimum L1-norm optimization model is constructed, and the gradient projection for sparse method is selected to solve the problem. Subsequently, in order to verify the performance of forward-looking imaging algorithm, one hardware-in-the-loop simulation system based on GPU is developed and the testing process is given. Then the hardware modules of the system are described in detail, mainly including simulation and control platform, echo signal generation platform and signal processing platform based on GPU. Aim at the implementation of forward-looking imaging algorithm on signal processing platform. By the analysis of its parallelism, the division of the functional modules is completed. Next, the parallelization of forward-looking imaging algorithm is realized with the use of the Compute Unified Device Architecture(CUDA) libraries and optimal strategy. More importantly, a graphical user interface is designed. Then the parameters transfer and algorithmic control operation can be done by means of it. Finally, the test and evaluation of the hardware-in-the-loop simulation system are completed. By comparing the operation period of algorithm on CUDA and C platform, the results show that GPU achieves better performance than CPU when the computation is equal.