Robust delay estimation for internet multimedia applications

Traffic estimation and modeling has been a crucial issue in many research areas of communication networks. For example, intermediate routers in networks estimate the rate of packet flows via queue size information in order to maximize throughput and provide fairness between flows. Also, many senders in the end system such as multimedia applications estimate end-to-end delay between a sender and a receiver to reduce dropping probability of packet. In this position paper, we propose a robust adaptive estimation scheme for Internet multimedia applications. The proposed scheme adopts an autoregressive (AR) model for the process and identifies the parameters of the AR process via a robust identification algorithm. This robust identification algorithm usually leads to better performance when the noise is correlated and/or nonstationary, and it is also more robust to modeling uncertainties. Here, we consider the problem of end-to-end delay estimation for audio playout mechanisms. We rigorously formulate the proposed scheme in the realm of audio playout mechanisms and give some preliminary simulation result which shows effectiveness of the proposed scheme. Even though we apply the proposed scheme to the estimation of end-to-end delay in audio applications, the scheme itself is very general and we expect that it can be applied to many other estimation problems in Internet multimedia applications.