Software-based Design And Realization Of The Radar Based On The Blade Server

With the development of hardware and blade servers,in order to achieve the expansion of radar functions and the reconstruction of software,we build a processing framework in the form of software and realize radar signal processing in the form of modularization for dynamic functional requirements.That is to say,changing the model of "the specific application requirements oriented" into the model of "the dynamic application requirements oriented".GPU has been used to speed up general computing because of its powerful computing power,high memory bandwidth and multi-threading characteristics in the field of general computing.As well as the appearance of Array Fire library with high performance and parallelism,it realizes the heterogeneous parallel computing between CPU and GPU.Because of its rich interface,it achieves the effect of hardware transparency.This paper realizes the software processing of radar signal by using GPU and Array Fire Library under blade server,which brings many conveniences such as software definition,expansion and upgrade of system function.Firstly,this paper introduces the background and significance of this research,and discusses the research status and the trend of software radar development.The radar signal processing is studied according to the application requirements.Secondly,the characteristics of the blade server are introduced,and the data flow direction of the blade server is analyzed according to the radar signal processing flow.Then the framework of signal processing is built under the blade server,and the corresponding functional modules are established according to the functional requirements.Finally,each module is connected under the framework to achieve the final application requirements.The main tasks are as follows:(1)Analyzing the radar signal processing and the interaction of platform and data according to the application scenario of radar.This part provides the platform foundation for the following work.(2)Learning the structure and usage of blade server,the theory and usage of the matrix operation function library of Array Fire,and the data transmission mode of communication middleware between nodes.The GPU acceleration characteristics and data interaction mode under heterogeneous computing are studied,and the processing flow is combined with blade server.The data flow trend of radar computing tasks under blade server,such as computing task allocation and monitoring parameters are analyzed.This part ensures the reasonable division of the signal processing work.(3)The parallel scheme of the system is analyzed,the framework of the application software is built based on the blade server and Array Fire.The task module is compiled under the framework,and the radar signal processing is realized by embedded model in series.Finally,the final operation results are analyzed and optimized.In this thesis,the software processing function of radar signal processing based on blade server is studied,which is the focus of this paper.This method breaks the conventional hardware binding mode,and can develop a scalable and reconfigurable radar signal processing terminal in a dynamical,inexpensive and efficient way according to mission requirements,which verify the feasibility of the parallel scheme.