MMC05-3: Distortion Estimation Algorithms (DEAs) for Wireless Video Streaming

This paper proposes three methods to estimate the distortion deriving from packet losses in wireless video communication. The proposed models take into account the short-term properties of the encoded video sequences and the properties of the wireless links. A suitable set of functions is adopted to model the distortion envelope resulting from multiple losses. The performance of the employed models, which depends on the average packet loss rate (PLR), is compared with the actual distortion, evaluated by decoding the received sequence with a properly designed, robust decoder. Numerical results confirm the accuracy of the proposed models, that are able to closely approximate the actual mean square error (MSE) behavior for a wide range of loss rates. Some applications of the proposed algorithms are also presented in the paper.