Software-defined Radar Signal Processing System Design And Realization Based On The Blade Server

With the development of radar technology,the focus of modern radar research and development has changed from "for specific special functions" to "for actual dynamic needs." Due to the excellent resource scheduling and task management capabilities of the CPU,the superior computing performance and storage bandwidth capabilities of the GPU,their combination is often used for accelerating computing in a variety of fields.In addition,the blade server is usually used as infrastructure platforms for all kinds of hardware because its advantages of high performance,low cost,high integration,and ease of management.With the emergence of computing middleware and communication middleware based on CPU,GPU,and blade server platforms,the decoupling of upper software and lower hardware has been completed,thus providing a possibility for the development model of software-based systems.Under the background of software radar,this thesis combines with the engineering project,the software design and implementation of the radar signal processing system are completed by using CPU and GPU on the blade server platform,the system's functions of easy scalability,high versatility,strong real-time,high portability and other performance are realized.This thesis mainly includes the following three aspects:1.Firstly,this thesis researches OpenMP programming model,CUDA programming model,the data interaction model of CPU and GPU,and the parallel optimization methods of CPU and GPU are analyzed.Then,the computing middleware technology and communication middleware technology are introduced,the relationship between middleware and underlying hardware is analyzed.Finally,the radar signal processing algorithms such as pulse compression,Doppler processing,and constant false alarm rate detection are studied in theoretical technology,implementation principle and simulation analysis.2.According to the project requirements,this thesis designs the architecture,functions,optimization,and data flow of the system.Then,the signal processing system is detailed decomposed and functionally designed according to the structural functions,the system is subdivided into scheduling and control module,signal processing module,and data sending module.Through the underlying programming model,middleware technology and signal processing technology,the functions of each module are realized on the blade server,which completes the unification of module interfaces and the allocation of system resources.Finally,according to the reliability and security requirements of the system,a series of design solutions are proposed.3.The test environment and simulation data are used to complete the test and analysis of the signal processing system's modules,which verifies each module's correctness,and the system's reliability and security.Besides,the performance test and optimization of the system are completed,which meets the real-time requirements of the system.