Quality Differentiation with Source Shaping and Forward Error Correction

The transmission of video traffic over the Internet is a fundamental issue of network design. Video applications require quality of service guarantees from the network in terms of limited packet loss, end-to-end delay, and delay variation. The question of today’s research and development is how to provide these guarantees considering the architecture of the present Internet. In the last years a variety of admission control schemes based on per-hop or end-to-end measurements has been suggested to control delay and loss sensitive streams with very little or no support at the routers. Most of these solutions, however, have to apply the same acceptance threshold for all streams, a significant limitation considering the diverse quality requirements of the applications. In this work we investigate how source shaping and forward error correction (FEC) can be used together to achieve application specific quality differentiation in terms of end-toend delay and packet loss probability. While source shaping and FEC have been proposed independently to decrease the probability of packet loss due to buffer overflow, their joint use has not been studied before. As the two control functions use the same scarce resource, end-node delay, and their efficiency to decrease loss probability is proportional to the introduced delay but with a decreasing marginal gain, combining the two a better performance can be achieved than by using only one of them. The performance evaluation focuses on the optimal delay allocation for shaping and FEC, such that the loss probability is minimized. We investigate how shaping can be used to substitute FEC redundancy and the sensitivity of the quality differentiation to the background traffic characteristics.