Systolic Array Implementation of DFT with Reduced Multipliers Using Triple Matrix Product

  • I. Mamatha
  • Shikha Tripathi
  • T. S. B. Sudarshan
  • Nikhil Bhattar
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 264)


A generic 2D systolic array for N point Discrete Fourier Transform using triple matrix product algorithm is proposed. The array can be used for a non power of two sized N point DFT where N=N1N2 is a composite number. It uses an array of size N2×(N1+1) which requires(2N+4N2) multipliers. For a DFT of size 4N2 (i.e multiple of four), an optimized design which requires 4N2 number of multipliers is proposed. It is observed that the proposed optimized structure reduces the number of multipliers by 66.6% as compared to the generic array structure while maintaining the same time complexity. Two examples are illustrated, one with non power of two size DFT and another with a DFT of size 4N2. Both the generic and optimized structures use the triple matrix product representation of DFT. The two structures are synthesized using Xilinx ISE 11.1 using the target device as xc5vtx240t-2ff1759 Virtex-5 FPGA. The proposed structure produces unscrambled stream of DFT sequence at output avoiding a necessity of reordering buffer. The array can be used for matrix -matrix multiplication and to compute the diagonal elements of a triple-matrix multiplication.


Discrete Fourier Transform Systolic Array FPGA 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Nandi, A., Patil, S.: Performance Evaluation of One Dimensional Systolic Array for FFT Processor. In: IEEE-ICSCN, pp. 168–171 (February 2007)Google Scholar
  2. 2.
    Cheng, C., Parhi, K.K.: Low-Cost Fast VLSI Algorithm for Discrete Fourier Transform. IEEE Transactions on Circuits and Systems-I: Regular Papers 54(4), 791–806 (2007)CrossRefMathSciNetGoogle Scholar
  3. 3.
    Kung, H.T.: Why Systolic Architectures? IEEE Computer 15(1), 37–46 (1982)CrossRefGoogle Scholar
  4. 4.
    Cooley, J.W., Tukey, J.W.: An Algorithm for the Machine Computation of Comlpex Fourier Series. Math.Comp. 19, 297–301 (1965)CrossRefMATHMathSciNetGoogle Scholar
  5. 5.
    Aravena, J.L.: Triple Matrix Product Architecture For Fast Signal Processing. IEEE Trans. Circuits Syst. 35(1), 119–122 (1988)CrossRefGoogle Scholar
  6. 6.
    Nash, J.G.: Computationally Efficient Systolic Architecture For Computing The Discrete Fourier Transform. IEEE Tran. Signal Processing 53(12), 4640–4651 (2005)CrossRefMathSciNetGoogle Scholar
  7. 7.
    Meher, P.K.: Highly Concurrent Reduced Complexity 2-D Systolic Array For Discrete Fourier Transform. IEEE Signal Proc. Letters 13(8), 481–484 (2006)CrossRefGoogle Scholar
  8. 8.
    Meher, P.K.: Efficient Systolic Implementation of DFT Using a Low- Complexity Convolution –Like Formulation. IEEE Transactions on Circuits and Systems-II: Express Briefs 53(8), 702–706 (2006)CrossRefMathSciNetGoogle Scholar
  9. 9.
    Meher, P.K., Patra, J.C., Vinod, A.P.: Efficient Systolic Designs for 1 and 2-Dimensional DFT of General Transform –Lengths for High Speed Wireless Communication Applications. Journal of Signal Processing Systems 60(1), 1–14 (2010)CrossRefGoogle Scholar
  10. 10.
    Chang, Y.N., Parhi, K.K.: An efficient pipeline FFT architecture. IEEE Transactions on Circuits and Systems-II:Analog and Digital Signal Processing 50(6), 322–325 (2003)CrossRefGoogle Scholar
  11. 11.
    Fan, X.: A VLSI-Oriented FFT Algorithm and Its Pipelined Design. In: ICSP 2008 Proceedings, pp. 414–417 (2008)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • I. Mamatha
    • 1
  • Shikha Tripathi
    • 1
  • T. S. B. Sudarshan
    • 1
  • Nikhil Bhattar
    • 2
  1. 1.Amrita School of EngineeringAmrita Vishwa VidyapeethamBengaluruIndia
  2. 2.Engineer Staff-II, IC DesignBroadcom India Research Pvt.LtdBangaloreIndia

Personalised recommendations