基于GPU的宽带信号时延差与相位差估计方法

针对多天线信号合成系统对于宽带、高速、并行信号的实时合成需求，设计了基于图形处理单元（Graphics Processing Unit，GPU）的宽带信号时延差与相位差估计方法，对估计方法中的快速傅里叶变换（Fast Fourier Transform,FFT)、共轭相乘、累加平均等模块进行了相应的线程并行程序设计。为充分发挥GPU的并行运算能力，利用异步流并发的方式对估计方法进一步优化，从结构层面有效提高了数据的并行处理效率。对基于GPU的宽带信号时延差与相位差估计方法进行了实验验证，多次实验测试结果表明，数据量为512 000时，在保证估计正确性的基础上，该方法相比传统串行CPU估计方法约有125倍的加速比，采用该方法可实现对多天线信号参数的实时估计。

Estimation Method of Delay and Phase of Wideband Signal Based on GPU

According to the real-time synthesis requirements of multi-antenna signal synthesis systems for wideband,high-speed,and parallel signals,an estimation method of delay and phase of wideband signal based on graphics processing unit(GPU) is designed.Thread-parallel optimization on modules such as fast Fourier transform(FFT),conjugate multiplication,and cumulative averaging,is performed.In order to give full play to the parallel computing capability of GPU,the estimation method is further optimized by using asynchronous stream concurrency,which effectively improves the parallel processing efficiency of data from the structural level.The proposed method is experimentally verified and the results show that when the amount of data is 512 000,on the basis of ensuring the correctness of the estimation,the method has a speedup of about 125 times compared with the traditional serial central processing unit(CPU) estimation method.The real-time estimation of multi-antenna signal parameters can be achieved by using the estimation method.