Wireless transmission of packet audio using multiple descriptions

We consider a scenario where a mobile user requests a high rate audio stream from the Internet via a wireless communications system. The congestion on the network and the instability of the wireless channel result in a loss of audio packets, which can impose a severe degradation on the high quality audio signal. To overcome this problem, we propose the use of multiple descriptions (MD). Our novel MD audio codec generates two complementary audio streams, which are sent in distinct packets to obtain the largest possible diversity gain. Preliminary simulation results support the flexibility and robustness of the proposed solution.