Error-resilient video transmission using long-term memorymotion-compensated prediction

摘要

Long-term memory prediction extends the spatial displacementnvector utilized in hybrid video coding by a variable time delay,npermitting the use of more than one reference frame for motionncompensation. This extension leads to improved rate-distortionnperformance. However, motion compensation in combination withntransmission errors leads to temporal error propagation that occurs whennthe reference frames at the coder and decoder differ. In this paper, wenpresent a framework that incorporates an estimated error intonrate-constrained motion estimation and mode decision. Experimentalnresults with a Rayleigh fading channel show that long-term memorynprediction significantly outperforms the single-frame predictionnH.263-based anchor. When a feedback channel is available, the decoderncan inform the encoder about successful or unsuccessful transmissionnevents by sending positive (ACK) or negative (NACK) acknowledgments.nThis information is utilized for updating the error estimates at thenencoder. Similar concepts, such as the ACK and NACK mode known from thenH.263 standard, are unified into a general framework providing superiorntransmission performance