End-to-end loss differentiation for video streaming with wireless link errors

摘要

Video delivery in heterogeneous wired/wireless networks is challenging, since link errors commonly degrade throughput performance, smoothness, and eventually impair the playback quality. We present an end-to-end Loss Differentiation Mechanism (LDA) which effectively decouples congestion from wireless loss in order to abolish the damage of error-induced multiplicative decrease on flow throughput and smoothness. The proposed LDA relies on Round Trip Time measurements to estimate the queuing delay and determine the appropriate error-recovery strategy. This mechanism can be easily adapted and incorporated into existing Additive Increase Multiplicative Decrease (AIMD) protocols, enabling them to utilize the available resources more efficiently. In this context, we incorporate the LDA into AIMD-based Scalable Streaming Video Protocol (SSVP), an end-to-end TCP-friendly protocol optimized for video streaming applications. Based on simulation results, we show that the combined approach provides the desired functionality to bind operationally wired and wireless links, within the framework of bandwidth efficiency, smoothness, and fairness.