Separation of chaotic signal sum into components in the presence of noise

The problem of separation of an observed sum of chaotic signals into the individual components is considered in the presence of noise. A noise threshold is found above which high-quality separation is impossible. This effect is shown to be associated with the information content of chaotic signals and a theoretical estimate is given for the threshold. A method for signal separation is proposed, which uses iteration of the chaotic source equations in reverse time. The method allows us to approach the theoretical limit threshold.