Distributed Radar System Target Detection And GPU Implementation

Large-scale ground-based distributed radar is a radar system that arranges multiple nodes in a large area and performs control and data processing through multiple process-ing centers.The radar of this system usually has high flexibility,strong detection ability,and wide detection range.,Anti-interference and anti-interception capabilities are strong.This thesis focuses on the research related to the target detection technology and the cor-responding software implementation of the distributed radar system.The main research contents are as follows:1.The loss of system coherent performance caused by time synchronization error,phase synchronization error and amplitude and phase error is analyzed.Aiming at the amplitude and phase error of the system,an algorithm based on least squares,combined with the MUSIC--like algorithm and the characteristics of distributed radar aperture ex-pansion,is used to estimate the closed-form solution of the amplitude and phase error of the transceiver array.Simulation experiments show that the method is effective.feasible.Aiming at the clutter and interference problems that may be encountered in the experimen-tal environment,a sum-and-difference network based on the least mean square algorithm is used to achieve interference suppression.Experiments show that the algorithm can suppress suppressive interference from different directions.2.Aiming at the problem of distributed radar receiving coherence,this paper studies a method of searching the airspace and calculating the coherent parameters of each search area,and then performing target detection after coherent synthesis of these areas.This method It solves the problem that echoes cannot be directly synthesized across distance units in different channels.The simulation experiment verifies the effectiveness of the method,and the engineering feasibility is verified through the experiment of measured data.3.Implemented GPU-based parallel processing of radar signals,including single-node signal processing modules and coherent synthesis modules.The structure changes of the input and output data in each module,the arrangement of the thread-block-grid structure in the GPU,and the data index corresponding to the operations performed in each thread are discussed in detail.Then the software running process in different modes is given,including single-machine single-GPU mode,single-machine multi-GPU and multi-machine multi-GPU mode.