Hybrid FEC and MDC models for low-delay packet-loss recovery
Real-time audio-visual communication requires a packet-loss recovery mechanism due to inevitable packet loss characteristics of the transmission channel such as the Internet. Especially for interactive applications e.g. real-time audio-visual communication, the delay for the recovery and enhancement technology must be minimized. Among the low-delay packet-loss recovery techniques, Multiple Description coding (MDC) and Forward Error Correction (FEC) are widely used for their robustness and efficiency. This paper proposes three new MDC-FEC models to minimize the overall distortion subject to specific source coding rates and channel packet loss probabilities. The proposed solution would enable adaptive selection of an optimal packet-loss recovery mechanism based on real-time feedback of channel conditions. The results indicate that this approach leads to an optimal model for joint source-channel coding based on a combination of MDC and FEC.