Binary arithmetic coding for time-varying sources based on the Maskit boundary

In this paper we propose a binary arithmetic coding technique for encoding the output of chaotic sources. The model we use for such sources is the free motion of particles on a surface of constant negative curvature. The source symbol probabilities are time-varying and follow a well defined rule as the one associated with a fractal known as Maskit boundary. We show that the average codeword lengths resulting from the proposed technique is less than the length of the original sequences.