Simple and reliable low power image communication based on DPCM and multiple refinements through feedback

There are basically two ways of reducing the power consumption in e.g. sensor nodes: the node algorithms should be simple and efficiently implemented, and the transmit power should be limited. In this paper we consider discrete time, progressive signal transmission with feedback. For the Gaussian case with an ideal return channel, the system performs according to OPTA (Berger, 1971). With substantial bandwidth expansion through multiple iterations, the power can be lowered close to a theoretical minimum. For a non-perfect return channel the performance is limited by the feedback channel's signal-to-noise ratio. We propose an image transmission system based on differential coding and multiple transmissions. The pixel predictions are done in the receiver to obtain a closed loop joint-source channel coding system, and to save processing power in the transmit node. Performance limits as well as simulation results will be given. The suggested system performs without delay.