SCIENTIA SINICA Informationis, Volume 47, Issue 7: 928(2017) https://doi.org/10.1360/N112016-00201

Direct acquisition method of long PN-code based on the GPU

More info
  • ReceivedAug 22, 2016
  • AcceptedJan 20, 2017
  • PublishedJul 6, 2017


In order to solve the acquisition problem of long PN-code spread spectrum signals, the partly correlation code phase-space acquisition model and implementation method of the FFT are analyzed. In order to segment data processing tasks and make full use of GPU to speed up FFT, a partly overlapped segmentation acquisition method of long PN-code is researched. Implementations of a direct global searching method and a partly overlapped segmentation searching method based on the GPU are then proposed, which parallel search for code phase using the FFT method, and accelerate FFT using the GPU. Simulation results indicate that, compared to a CPU implementation method, the search speed of the direct acquisition method of long PN-code based on the GPU is improved significantly.

Funded by


  • Figure 1

    Part correlation code phase space searching method

  • Figure 2

    Partly overlap segmentation and correlation of input samples

  • Figure 3

    Performance of coherent detection

  • Figure 4

    Gain of non-coherent accumulation

  • Figure 5

    Code phase acquisition results based on GPU. (a) Samples offset: 15000001, Doppler frequency: 0 Hz; protectłinebreak (b) samples offset: 15000001, Doppler frequency: 5 kHz; (c) zoom in (a) at 15000000 samples offset

  • Figure 6

    Doppler frequency estimation based on segment correlation

  • Table 1   Time consuming of implementation based on CPU and GPU
    Direct global
    searching algorithm
    Direct global
    searching algorithm
    Time consuming of CPU implementation (ms) 17600.3 4759.3
    Time consuming of GPU implementation (ms) 322.4 285.3
    Acceleration rate 54.6 16.7
  • Table 2   Each stage time consuming of direct acquisition algorithm based on GPU
    Stage of algorithm
    Direct global
    searching algorithm (ms)
    Direct global
    searching algorithm (ms)
    Zero padding and overlapping 19.6 44.9
    FFT of input signal 84.3 72.6
    FFT of local signal 84.2 0.1
    Conjugate multiplication 32.8 71.4
    IFFT 84.6 73.0
    Modulus 14.3 20.7
    Signal detection 2.6 2.6
    Total time consuming 322.4 285.3

Copyright 2019 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有