• DocumentCode
    2170617
  • Title

    Implementation of a digital down converter using graphics processing unit

  • Author

    Xiao Ma ; Lixia Deng ; Yuping Zhao

  • Author_Institution
    Peking Univ., Beijing, China
  • fYear
    2013
  • fDate
    17-19 Nov. 2013
  • Firstpage
    655
  • Lastpage
    660
  • Abstract
    This paper presents a DDC (digital down converter) on NVIDA 580 GTX, which consists of a DDS (direct digital synthesizer), a CIC (cascade integrator comb) decimation filter and a FIR (finite impulse response) filter. The decimating factor of the CIC decimation filter can be arbitrary positive integer and the major concern is concentrated on how to drive it to work well while the decimating factor varies. In our strategy, the problem is ranged into two cases according to whether or not the decimating factor is smaller than 128. Then, we provide suitable method to deal with data in each case. Additionally, we present different ways to construct a root raised cosine filter with 4 times oversampling on GPU (graphics process unit). Through flexible threads assignment and efficient scheduling strategy, the GPU-based DDC is implemented significantly. We evaluate the performance of the designed DDC with respect to its counterpart based on CPU (central processing unit) developed in the C language. Experimental results demonstrate that the DDC shows significant improvements on GPU and achieves a speedup of 387 times.
  • Keywords
    FIR filters; comb filters; digital signal processing chips; direct digital synthesis; frequency convertors; graphics processing units; multi-threading; parallel architectures; performance evaluation; scheduling; C language; CIC decimation filter; CPU; DDC performance evaluation; DDS; FIR filter; GPU-based DDC; NVIDA 580 GTX; arbitrary positive integer; cascade integrator comb decimation filter; central processing unit; decimating factor; digital down converter; direct digital synthesizer; finite impulse response filter; flexible threads assignment; graphics processing unit; root raised cosine filter; scheduling strategy; Central Processing Unit; Convolution; Equations; Finite impulse response filters; Graphics processing units; Instruction sets; Throughput; Graphics processing unit; cascade integrator comb; digital down converter;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Communication Technology (ICCT), 2013 15th IEEE International Conference on
  • Conference_Location
    Guilin
  • Type

    conf

  • DOI
    10.1109/ICCT.2013.6820456
  • Filename
    6820456