Adaptive video-aware forward error correction code allocation for reliable video transmission

The Internet is rapidly becoming the dominant platform for video contribution and distribution. As Internet provides a best-effort service without any guarantee of quality. The video streaming has to cope with lack of guarantees. Due to congestion and the heterogeneous nature of Internet, a video transmission will be plagued by packet loss, variability in throughput, and latency. To mitigate the aforementioned problems, forward error correction (FEC) approach can be employed. Among different FEC schemes, due to the maximum distance separable property, Reed–Solomon (RS) erasure codes are commonly used as a method to protect the video streams when transmitted over unreliable networks. As a block-based error-correcting code, smaller block size decreases the RS code efficacy and decoding delay and larger block size increase decoding delay with the enhanced RS coding performance. Moreover, on video data, the RS codes are typically applied at the packet, frame, and sub-GoP level. In this paper, we provide a comparative analysis of different FEC approaches used to ensure reliable video transmission and study the relationship of all the approaches for different network scenarios. We presented a decision tree-based machine learning model to apply FEC by considering network conditions, video encoding parameters, and motion characteristics. With the achieved accuracy of 81%, our proposed model applied FEC on video data while maintaining the required quality of experience (QoE). The benefits of the proposed mechanism are demonstrated by carefully designed experiments and assessed through objective QoE metric.