Distributed Audio Transcoding For Peer-to-Peer Systems

Mobile devices often connect to the network via wireless channels. In general, the downlink of the wireless channel (e.g., the cellular access network) is limited in throughput. The wireless channel may also suffer from transient fading or unexpected interferences, which results in throughput fluctuation. Thus, it is desirable to adjust the bitrate of the audio stream over time to adapt to channel conditions. In addition, it is desirable if one can adapt the audio signal to individual mobile users depending on the requirements of applications and hardware limitations of mobile devices (e.g., low-end built-in speaker). The change in the bitrate or the quality of the audio signal can be achieved by transcoding.

This invention describes a distributed transcoding algorithm for audio streaming to mobile users in P2P systems. The algorithm allows several fixed nodes to perform transcoding for a mobile device. After a mobile device connects to multiple fixed nodes as its parents, each parent generates a substream by transcoding the original audio. These substreams are transmitted and assembled at the mobile device as if they were a single stream. If the mobile device loses some of its parents, it can decode the incomplete signal and the user experiences graceful degradation instead of complete disruption.