GPU-based fast processing for a distributed acoustic sensor using an LFM pulse

We carried out a fast processing investigation based on a graphics processing unit (GPU) for a distributed acoustic sensor using a linear frequency modulation pulse. The moving window cross-correlation calculations are realized on the GPU, which makes use of parallel computing. We analyzed the effect of the thread number in a block on the GPU streaming multiprocessor utilization efficiency and then compared the acceleration under different calculation scales. By maximizing the streaming multiprocessor utilization efficiency and large calculation scale, a maximum acceleration ratio of 86.01 was obtained.