White Gaussian Noise Generator - 2020.2 English

Vivado Design Suite Reference Guide: Model-Based DSP Design Using System Generator (UG958)

Document ID
Release Date
2020.2 English

The Xilinx® White Gaussian Noise Generator (WGNG) generates white Gaussian noise using a combination of the Box-Muller algorithm and the Central Limit Theorem following the general approach described in [1] (reference listed below).

The Box-Muller algorithm generates a unit normal random variable using a transformation of two independent random variables that are uniformly distributed over [0,1]. This is accomplished by storing Box-Muller function values in ROMs and addressing them with uniform random variables.

The uniform random variables are produced by multiple-bit leap-forward LFSRs. A standard LFSR generates one output per clock cycle. K-bit leap-forward LFSRs are able to generate k outputs in a single cycle. For example, a 4-bit leap-forward LFSR outputs a discrete uniform random variable between 0 and 15. A portion of the 48-bit block parameter seed initializes each LFSR allowing customization. The outputs of four parallel Box-Muller Subsystems are averaged to obtain a probability density function (PDF) that is Gaussian to within 0.2% out to 4.8sigma. The overall latency of the WGNG is 10 clock cycles. The output port noise is a 12 bit signed number with 7 bits after the binary point.

4-bit Leap-Forward LFSR

Figure 1. 4-bit Leap-Forward LFSR Diagram

Box-Muller Method

Figure 2. Box-Muller Method Diagram

Block Parameters

The block parameters dialog box can be invoked by double-clicking the icon in your Simulink model.

The block parameter is a decimal starting seed value.


A. Ghazel, E. Boutillon, J. L. Danger, G. Gulak and H. Laamari, Design and Performance Analysis of a High Speed AWGN Communication Channel Emulator, IEEE PACRIM Conference, Victoria, B. C., Aug. 2001.