Frame-based forward error correction using content-dependent coding for video streaming applications

Abstract

Forward error correction (FEC) is a common error control technique to improve the quality of video streaming over unreliable channels by recovering packet losses during transmission. The traditional frame-level FEC adopts the independent FEC coding method, regarding each video frame as one coding block, and protects each block with its respective FEC data according to the principle of unequal error protection (UEP). In contrast, emerging content-dependent FEC coding schemes explore the performance improvement provided by the expandable block structure. For the frame-level FEC with an expandable block structure, a high-priority frame is involved in the FEC processing of its following frame, and thus the frames with a dependency relation in the video coding stage are also correlated in the FEC coding stage. In this paper, an analytical content-dependent FEC model is developed to estimate video streaming quality in terms of playable frame rate. Based on this model, optimized FEC configuration can be found given the inputs of the video and transmission channel parameters. A comprehensive performance study is conducted for the proposed model to observe the redundancy assignment behavior, the reconstructed video quality and the FEC coding complexity, under the various network situations. The performance results show that the expanding-block FEC can be superior in balancing the trade-off between FEC complexity and loss recovery capacity.