As the world becomes more connected and information flows rapidly between the billions
of devices, snooping of nodes or security risks of information transfer are increasing. As a result,
increasing levels of cryptography are required to keep private data safe and secure. Most
cryptographic techniques rely on obtaining random sequences as keys or ciphers to encode and
encrypt information. Dynamic environments lead to rapid establishment and teardown of sessions
in networked applications, requiring high rates of random sequences, and placing a burden on
random number generators (RNG) to deliver these sequences both quickly and securely. This
thesis presents a new type of entropy extractor to address these needs. The proposed
architecture introduces an architecture that uses a whitening filter in feedback to increase the
potential entropy extracted from the noise source. Due to its placement in the loop, the whitening
filter is implemented in the digital domain, allowing it to be reconfigured to work with a wide variety
of noise sources. This thesis shows how the architecture was developed and analyzes the effects
different component choices may have on the entropy extractor’s performance.