Abstract
Combining the Box-Muller method, Central Limit Theorem, and the Wallace method, a hardware white Gaussian noise generator (WGNG) is proposed to simulate the noise effect appeared in the communication channel and is synthesized in a 0.18um CMOS process. Passing two statistical tests of chi-square test and Kolmogorov-Smirnov (K-S) test, the proposed noise generator can generate high-quality 666.667 million Gaussian random variables per second.It is different from the existing methods that require complex calculations. The proposed design only requires additions, subtractions, and shift operation in the major part. Because of only having simple operations, it is easy to achieve high performance.In addition, the proposed architecture is not only applied to generate the additive white Gaussian noise (AWGN), but also applied to generate the random variables with other distributions such as exponential distributions.