Study Of Missile-borne Signal Processing Based On Multicore Processor

As the center of precision guidance in modern warfare,the missile-borne computer not only need to process the information such as the position and attitude of the missile in real time,but also complete the detecting,imaging,identifying and positioning operation of the target group,so the missile-borne computer has a high demand of processor's calculation performance and data transmission bandwidth.In the face of increasingly complex war scenes and multi-functional composite guidance,missile-borne computer based on traditional single-core processor has been unable to meet the demand.This thesis has studied the missile signal processing based on multi-core processor,including the design of multi-core processor system architecture and the reserch of key issues and technical in signal processing,the experimental result shows that through reasonable task mapping and resource allocation,multi-core processor can significantly exert their advantages to enhance the computing performance of missile-borne computer and meet the growing requirements of modern warfare.The main works of this thesis are as follows:Firstly,we choose SAR imaging algorithm as the main research contents,and give introduction of its basic principle,then we take the classical RD algorithm as example and discuss the signal model and engineering realization in multicore processor.Secondly,we set up two kinds of missile-borne signal processing platform,they are based on multicore DSP and Embedded GPU.Thirdly,the theoretical analysis and verification of the signal processing based on multi-core DSP are carried out.In addition to the underlying functions' optimization,inter-core synchronization and data communication,we also focus on the key technologies such as large-scale matrix data storage and very large FFT implementation.Finally,a set of measured radar echo data are utilized to verify the performance of multi-core DSP.Fourthly,in the aspect of missile signal processing software design based on GPU,a detailed demonstration of engineering implementation of two-dimensional imaging algorithm is given,and the Stream techniques are used to promote GPU's parallel processing capability.At the end of the paper,we compare the performance of the two kinds of multi-core processors,the result shows that GPU has obvious advantages in the intensive and high-parallell computing.