Software defined ultra-low latency Video-Over-IP system with compression

Speaker: Siegfried Foessel

• Traditional Video-over-IP implementations in software either result in higher latency due to the processing time for compression or in high bandwidth required when transmitted as an uncompressed stream. With 4k and 8k video, this is even more of a challenge, as the uncompressed stream requires high-performance Ethernet networks or the compression needs special hardware support. In some cases however, a software implementation and standard COTS equipment are beneficial to allow higher flexibility. With JPEG XS, a mezzanine compression codec was developed that can also be implemented as an ultra-low latency system in software. However, special care must be taken when designing a multi-threading architecture and an optimized data pipeline. Using a case study for 4k Video-over-IP, this paper explains how such a system can be implemented with COTS components, which software architecture is necessary and how low the latency can be reduced.

Performance of Low-Latency DASH/CMAF and Low-Latency HLS Streaming Systems

Speaker: Yuriy Reznik

• Reducing end-to-end streaming latency is critical to HTTP-based live video streaming. There are currently two technologies in this domain: Low-Latency HLS (LL-HLS) and Low-Latency DASH (LL-DASH). The latter is sometimes also referred to as Low-Latency CMAF (LL-CMAF), but effectively it is the same architecture. Many players support LL-HLS and/or LL-DASH protocols, including Apple's AVPlayer, Shaka player, HLS.js, DASH.js, and others. There are also several encoding and packaging tools available for the generation of LL-HLS and LL-DASH streams. This paper is dedicated to the analysis of the performance of low-latency players and streaming systems. The evaluation is based on a series of live streaming experiments, repeated using identical video content, encoders, encoding profiles, and network conditions, emulated by using traces of real-world networks. Several performance metrics, such as average stream bitrate, the amounts of downloaded media data, streaming latency, as well as buffering and stream switching statistics are captured and reported in our experiments. These results are subsequently used to describe the observed differences in the performance of LL-HLS and LL-DASH-based systems.