Toward a theory of secure communications in a non-random environment

An open problem is to quantify the probability of a signal from a noisy environment when little is known about its signal generating properties. The signals can have a mix of random, correlated, and deterministic elements. The only operating assumption is that the larger system needed to generate the signal with a high probability, the less likely the signal. This paper considers a class of models that can reproduce a mix of random, correlated, and deterministic signals depending on the value of the model´s parameters. The approach is to find the parameters that yield the maximum probability of generating a given signal. This maximum probability, because it uses the optimum parameters, is larger than the probability of generating the signal from a noisy environment.