Adaptive median thresholding for the generation of high-data-rate random-like unpredictable binary sequences with chaos

Chaos represents an effective method for the generation of random-like values, combining the benefits of relying on simple, causal models with good unpredictability. Since chaotic systems are always continuous-valued while applications are often digital, a quantization step is generally required for interfacing, We propose an adaptive binary quantization scheme which allows a relaxation of the system accuracy requirements, permitting improved performance in terms of throughput, die area, design simplicity.