基于嵌入式GPU的三天线GNSS基带信号处理加速技术

GNSS软件接收机因其可移植性及灵活性等优点,持续受到业内关注。但传统的软件接收机存在计算量大、耗时长的捕获和跟踪基带信号处理过程,使得软件接收机往往跟踪通道较少,且难以在嵌入式系统上运行。利用嵌入式图形处理器(GPU)的高浮点性能和并行运算能力,对GNSS软件接收机中耗时长且并行性明显的模块进行加速,实现采用CUDA流的数据读取,以及多采样点并行的捕获和多采样点、多卫星并行的跟踪。采用嵌入式GPU进行加速后,可将数据读取速度提高3.43倍,卫星搜捕速度提高16.83倍,卫星跟踪速度提高11.28倍。实验结果表明,在嵌入式Jetson TX2平台上可以支持超过90个62MHz采样的GNSS卫星信号处理。研制了三天线GNSS信号的定位和测姿接收机,为未来小型嵌入式PNT系统的研制提供新思路。

Signal Processing Acceleration Technology of Three-antenna GNSS Baseband Based on Embedded GPU

GNSS software receiver is being paid more and more attention by researchers due to its portability and flexibility. However, the process of capturing and tracking baseband signals in traditional software receiver is very computationally intensive and time-consuming, which makes it difficult to run on embedded systems. The time-consuming and parallel modules in the GNSS software receiver are accelerated using embedded GPU, which has high floating point performance and parallel computational power. The data reader module uses CUDA stream for acceleration, which can increase the data reading speed by 3.43 times. The acquisition module uses multiple sampling points in parallel for acceleration, increasing the acquisition speed by 16.83 times. The tracking module uses multiple sampling points and multiple satellites in parallel for acceleration,