DSP 9: Digital Noise Generation

(Note: this fell off - I was just having too much fun in Korea, then Thailand! But now I'm back in Hoi An, Vietnam, and feeling good to get back to this.)

Covered:

• Chapter 2: Digital Noise Generation

We often need to generate digital noise, which is a random signal that follows the Gaussian PDF. This noise is used to ensure that devices, signals, and algorithms work given the presence of random noise.

The Central Limit Theorem basically states that the distribution of random numbers drawn from independent random sources will converge to the Normal / Gaussian PDF. This is why the normal distribution gets so much attention - it allows statisticians to work with random processes with known Gaussian techniques, I suppose.

The below shows how the PDFs of different RND generation processes converges to the Gaussian - the final one is 12 RND sources and the PDF looks very Gaussian indeed:

Random number generators (RNGs) use a seed, often a timestamp, to generate random numbers by permuting the seed through a known algorithm to get the next random number. These make these numbers pseudo-random, since they're reproducible given a known seed.