Study for Randomness of Diverse Pseudo-Random Bit Generators Oriented to Telecommunication Applications

The paper presents the results from the research for randomness of 8 pseudo random bit generators (PRBGs) for short sequences (256, 512 and 1024 bits) - Linear Feedback Serial Registers (LFSRs) with 8, 9, 16, 21 bits and different taps, Galois, Kasami, Gold and Mersenne-Twister (used in MATLAB). The goal is to determine the best PRBGs for telecommunications implementations for which randomness is critical as Analog-to-Information Converters (AICs) based on compressive sensing, where the recovery of the signal samples at under-Nyquist rate uses a random bit sequence. The PRBGs are tested with 10 tests from NIST suite for randomness which are applicable for short sequences. The PRBGs passing the tests are implemented in a laboratory set-up of AIC for obtaining the solution with the highest Signal-to-Noise and Distortion (SINAD).