Video traffic accounts for more than 60% of overall traffic across practically any network in the world. This number is expected to grow, with Cisco predicting video will reach 82% of global IP traffic by 2022. This prediction makes sense, especially due to the recent slew of services and applications challenging the dominance of Netflix and Goggle’s YouTube. Those include social platforms such as TikTok and OTT services like Disney+, Apple TV+, AT&T/WarnerMedia’s HBO Max and Comcast/NBCUniversal’s Peacock.
With such dominance in the international internet traffic mix, it was very exciting to see that a new video codec standard – Versatile Video Coding (VVC) or H.266 – was finalized recently with a promise of reducing the capacity of video traffic by potentially 50%, with no quality loss. A similar promise was made by Qualcomm, describing a 40% increase in video streaming efficiency with the new codec.
Could this lead to a significant reduction in the demand for capacity in mobile and fixed networks? Does this mean we do not really need to upgrade our core, backhaul and access infrastructures to support much higher capacity demand?
The short answer is no.
There are several reasons for that:
- Video profiles are constantly growing: from standard-definition (SD) and high-definition (HD) to 4K and 8K ultra-high-definition (UHD) resolutions; from 30 frames per second (FPS) to 60, 90 and even 120 FPS; and from single views to 180-degree or 360-degree views. This means that while the new H.266 video standard is more efficient per resolution and frame rate, it probably will be utilized in use cases requiring higher resolution, eventually consuming more capacity. MPEG-2/H.262 needed 4Mbps for a typical (at the time) SD signal, while the more efficient H.264/AVC needed 8Mbps for a HD signal. Today, H.265/HEVC uses about 11Mbps for a 4K signal, and H.266/VVC will utilize around 14Mbps for an 8K signal.
- Using advanced video codecs is not without a price. There is a trade-off between the codec’s efficiency in terms of capacity, resources and time (and hence, latency) required for both coding and decoding. Advanced coding requires more processing power, which limits its implementation. It also adds more latency, as shown in a BBC R&D report indicating a 6.5x increase in encoding time and a 1.5x increase in decoding time for H.266 compared to H.265.
- Using advanced video codecs is not without a price (2). The H.266 is a royalty-based codec, as were the previous H.26X and MPEG-X codecs. But as those dominated the market for years, a significant step was made in the royalty-free codec domain with the formation of the Alliance of Open Media (AOMedia) and the introduction of the AV1 codec. This codec replaced the existing VP9 and planned VP10 codecs (from Google) as well as Cisco’s Thor and Mozilla’s Daala codecs. That made AV1 the de facto industry standard for non-royalty codecs, with a growing support base. It also demonstrates superior performance over the H.265/HEVC codec – IEEE found it to be 25% more efficient than H.265 for a Netflix-like traffic set. This codec is moving into Google’s YouTube and Netflix, which hold, according to the latest Sandvine report, a combined share of over 20% of global download traffic. This means that the H.265 benchmark quoted with regards to H.266 codec efficiency is not the right one. H.266 is only about 25% more efficient than AV1, which does not seem to be appealing enough when comparing a royalty-free codec to a royalty-based one.
So no, do not stop investing in your infrastructure. Video traffic will continue to dominate your traffic mix. While being more efficient, video traffic will continue to grow.
How will your network be affected by the video revolution?