基于GPU的GPS信号并行捕获

针对计算机中央处理器上串行实现GPS捕获算法耗时长的缺点,利用具有强并行处理能力的图形处理器设计实现了两种分别适用于不同载噪比信号的并行捕获算法以提高捕获速度。所提算法基于计算机统一设备架构的设计思想,采用了并行码相位搜索捕获策略,通过对GPS星座32颗卫星多通道、多频点的并行搜索实现了强信号捕获,而对弱信号则采用非相关积分法,通过对单颗卫星多时段、多频点的并行搜索再进行通道的串行处理来实现并行捕获。仿真结果表明:两种并行捕获算法比串行实现的捕获算法速度提高了10倍;采用非相干积分提高了弱信号捕获能力,对于载噪比为40 dB的10 ms中频数据,在保证捕获速度的同时,仍能够有效实现正确捕获。

Parallel-acquisition of GPS signal based on graphic processing unit

Since the serial realization of GPS acquisition algorithm on CPU of PC is time-consuming,a parallel realization of two acquisition algorithms on graphic processing unit(GPU) were proposed for the GPS signal with different carrier to noise ratio(CNR) to improve the calculation speed.These algorithms were designed according to compute unified device architecture(CUDA) based on the parallel code phase search technology.The strong signal acquisition with high CNR was realized by parallel search of multi-frequency for all 32 satellites.As for the weak signal acquisition with low CNR,the noncoherent integration was adopted.The experimental results show that the calculation speeds of the two proposed algorithms are about 10 times faster than that of the traditional method,and for the weak signal with 40 dB noise,the proposed parallel realization algorithm acquires satellites effectively with the similar calculation speed by using noncoherent integration of 10 ms data.