Adaptive video coding rate control and adaptive bit-rate allocation to source-channel coding for time-varying channels

Abstract

Multimedia applications such as video streaming, video-phone, video-conference are becoming the key services in the Internet and mobile networks. These networks, however, are very prone to packet loss or bit error, in addition to their time-varying characteristics in available channel rate and delay. So they can hardly support the necessary QoS (quality of service) for a video source to transmit its traffic without loss or violation of delay-constraints during a session. The typical method to control QoS in transporting video over time-varying channels is to use a QoS control scheme at the application-layer, which has two challenging issues; end-to-end flow control and error control. In this dissertation, we present a practical yet novel approach to respond to the above issues. Specifically we focus on a new adaptive video coding rate control scheme in a time-varying available bandwidth channel for an end-to-end flow control problem, and a code rate allocation scheme to a source-channel coding in a time-varying packet loss channel for an error control problem in application-layer QoS control. First, we propose a practical and accurate R-Q (rate-quantization) model and D-Q (distortion-quantization) model per picture frame with moderate complexity and relatively low processing delay for video rate control. These are to estimate the generated bits and distortions accurately in DCT(discrete cosine transform)-based coding such as H.263. The proposed R-Q model is based on a linear relationship between non-zero level count of the DCT coefficients and the generated bits, while the proposed D-Q model comes from the observation that the squared quantization parameter over distortion can be very accurately approximated by a quadratic function of quantization parameter. We verify that the proposed models estimate the real coding results very accurately, and can be effectively applied to video coding rate control. Second, we propose a new adaptive video rate control schem...