White Gaussian Noise Generator
Software estimation of the performances of a communication system is very time-consuming. Moreover, many variables (sampling frequency, digital format, carrier resolution, rounding and quantization etc.) have to be optimized for satisfying the best trade-off between performances and complexity. In order to speed up the final parameter optimization of a design, we propose to perform direct hardware simulation (emulation) on an FPGA. Such a simulation needs a hardware emulation of the communication channel.
To do so, we have designed a hardware White Gaussian Noise Generator (WGNG) in a FPGA
circuit. High accuracy, fast and low-cost hardware are reached by combining the
Box-Muller and Central limit methods. The design is fully parameterizable and the
performance of the WGNG is entirely caracterized with a MATLAB program.
This WGNG is also used to generate more complex channels, such as Rayleigh channel and
For more information, you can download the slides
of a talk given at the university of Toronto(Aug. 2001):
This work is the result of a joined research project carried out by :
- LESTER (France), with Emmanuel Boutillon
- ENST-Paris (France) with Jean-Luc Danger
- SUP’COM (Tunisia), with Adel Gazel
- University of Toronto (Canada), with Glenn Gulak
- E. Boutillon, J.L. Danger, A. Gazel, "
Design of High Speed AWGN Communication Channel Emulator" Kluwer Press,
Analog Integrated Circuits and Signal Processing 34(2): 133-142; Feb 2003
- D. Derrien, E. Boutillon,"
"Quality Measurement of a Colored Gaussian Noise Generator Hardware
Implementation Based on Statistical Properties", IEEE International
Symposium on Signal Processing and Information Technology, Maroc, Déc. 2002.
- E. Boutillon, D. Derrien, "Implémentation hardware
d’un générateur de bruit de Rayleigh coloré",
Journées Francophones sur l’Adéquation Algorithme Architecture, Tunisie, Déc. 2002. .
- A. Gazel, E. Boutillon, J.L. Danger, G. Gulak, H. Lamaari,"Design and performance analysis of a high speed AWGN
Communication Channel Emulator" (PACRIM'01), Victoria, British Colombia, Canada, August 2001
- J.L Danger, A. Ghazel, E. Boutillon H. Laamari,"Efficient FPGA Implementation of Gaussian Noise Generator for Communication Channel Emulation", The 7th
IEEE International Conference on Electronicsm Circuits & Systemes (ICECS'2K),
, Kaslik, Lebanon, Dec 2000
E. Boutillon, A. Gazel, J.L. Danger, G. Gulak, H. Laamari, "
Un générateur de bruit blanc gaussien sur un FPGA pour la simulation rapide de systèmes de transmissions", 14ième Colloque du GRETSI, Toulouse Sept. 2001
The MATLAB source file can be directly downloaded:
The VHDL source files are free under a GPL licence, please,
send an email at firstname.lastname@example.org with the subjet
"request for WGNG source files" and you will receive the source files by email
(so far, you have been more than 200 hundred to download the files).
- Method using 2's complement
- Method using 1's complement (as in VHDL files)
All you comments and improvements on this WEB site and the source files are welcom.
Generic Hardware Random Generator
With Cedric Marchand (a former Ms. Degree), we have improved and made a VHDL implementation
of the very efficient technic of  and more recently, the technic proposed by David B.
Thomas and Wayne Luk  to perform a generic random generator.
This new generic random generator has the following features:
If you have question about this new architecture (matlab and VHDL files),
please, send an email at email@example.com with the subjet
"Question about the Generic Hardware Random Generator"
 A.J. Walker, "An efficient method for generation discrete random varialbes with general distribution",
ACM, Trans. Math. Software 3, pp 253-56, 1977.
 Non-Uniform Random Number Generation Through Piecewise Linear Approximations, "International Conference onField Programmable Logic and Applications, 2006. FPL '06.
, pp 1-6, Aug. 2006.
- FPGA implementation with a very low hardware cost;
- it can emulate every type of continuous distributions with the desire precision;
- it has also capabilities to emulate non continuous distribution (P(X=x) = \alpha * exp(-x), x > 0, P(X=x) = 1/sqrt(1-x^2), -1
This section gives more information on WGNG generators
XILINX IP core
- COMBLOCK IP core
A gaussian noise generator for hardware based simulation", D. Lee et al, IEEE trans. on computer, vol 53, n° 12,dec. 04.
- D. Lee homepage
Return to main page of Emmanuel Boutillon
Return to the main page of the lab
Return to the main page of the university