Rate Control for a Video Coder using Learning Automata

In this paper, a rate controller for a H.261 based video encoder is proposed. The rate controller adaptively chooses the optimal channel matched quantizer using a stochastic learning automaton. The automaton learns the channel characteristics based on a one bit feedback from the decoder. The rate control algorithm is shown to converge to the optimal choice of the quantizer very quickly for various channel bit error probabilities and for different video sequences. The adaptation can be achieved in real-time. The peak signal to noise ratio of the received video signal is seen to be better using the proposed approach.