Global ICT Standardization Forum for India (GISFI)
Title: Greening the network with video traffic control
Company: NEC Corporation
Purpose: Discussion and Approval
Doc number: GE-20110008
Meeting: GISFI#5, Hyderabad, India, 20 Jun – 22 Jun, 2011
1. Abstract
The worldwide trend of the internet is moving towards video contents. As it can be easily assumed, rich video contents can use large bandwidths of communication systems and will cause pain to internet service providers. Linear increase of equipment with the growth of the needed bandwidth is one solution to this problem, but will also cause impacts to the earth’s environment at the same time. This can also be said about India, where IT usage is growing at a very fast pace. Effectively controlling the contents at the user plane level can increase user experience and at the same time avoid unneeded expansion of the network, resulting in a “green” network. We propose this document to be accepted for the deliverable 2 of Green Energy activity, “Study on potential enhancements of ICT (Information and Communications Technology).”
2. Introduction
Video data is one section of internet traffic that is growing at a tremendous pace. Service providers will need to reinforce their network capacities in order to prevent traffic from overflowing and cause impact to user experiences. According to the Cisco Visual Networking Index [1], it is said that internet video traffic will grow to be 57% of all internet traffic by 2014. This may not be the exact figure for India, but the similar phenomenon can be easily predicted.
Overflow problems of the network can be easily solved by adding network equipment to deal with the increased traffic. However, more equipment would mean more power consumption, which results in more Green House Gas (GHG) emission. Alternative methods to secure user experience and lighten the burden of the networks are needed to realize a greener network.
In this document, we will first discuss the future trends of internet traffic. Second, we will discuss current solutions available and perform a gap analysis of what elements are missing. Finally, we will propose a high level solution to cope with the problem.
3. Current Analysis
According to the Cisco Visual Networking Index [1], worldwide internet video will grow at a CAGR (Compound Annual Growth Rate) of 48% between 2009 and 2014. As for the Indian internet population, it is stated to grow from 14.6 million people to 42.1 million people from 2009 to 2014 (CAGR: 23%). This is one of the highest growth rates in the world but is only a small portion of the Indian population. Therefore, it can be easily assumed that the growth will still continue even after 2014 and since its user growth is faster than the worldwide average, it will have a faster growth than CAGR 48%.
A growing population of internet video users will mean more personal equipment at the user end and more equipment in between the user and service provider. This results in more energy consumption and more GHG emissions.
3.1. Known solutions
An effective method to reduce bandwidth to enable more users would be to compress the video bandwidth by using a codec with a higher compression rate. For example, using the MPEG-4/H.264 codec would have twice as better compression rates compared to MPEG-2 [2]. If the communication endpoints do not support the codec, it is possible to place Media Gateways (MGW) in necessary areas of the network. The MGW gateways may be able to output video optimized for each user (e.g. sending smaller sized videos compared to fullHD size videos to mobile phones with small screens).
However, since higher compression rates would require intense calculations, changing the codec type (especially in the middle of the network) may consume more energy than the original network. Furthermore, due to the intense calculations, the capacity of MGWs is relatively small and may end up with placing even more equipments. This will not lead to a green network and other solutions are needed.
Figure 1. Conventional methods for efficient network use
3.2. Gap analysis
Currently known solution depicted above may not result in a green network and other approaches are needed. The requirement for watching video would be to secure enough bandwidth so the user experience would not be greatly damaged. The bottleneck found in the last section was the massive calculation needed to change codec. Therefore, a method which can bypass this bottleneck is also a requirement.
4. High level proposals
In order to improve user experience and not increase GHG emission at the same time, a solution that does not involve complex calculations is desired. When multiple users exceeding the network capacity access the network, most of the user will not be able to see the video content at the desired user experience (e.g. the video will frequently halt, will take time to start, or will not play at all). Solutions that can solve the problems are needed.
4.1. Prioritization Solution
In order to meet both requirements, one solution would be to dynamically prioritize each user. One reason that user experiences degrade would be because all users have the same priority and clog the network. If some of the users’ priority level is degraded for a given time, the network would be able secure enough bandwidth to cover other active users. Once the download/upload of other users is finished, the degraded users’ priority will be put to normal and services will resume. In other words, this solution put a few users in the “victim” list to save the bandwidth of other users.
Who to put inside the victim list is of great importance. To give equal user experience to all of the network’s users, users who have been recently put in the victim list will be the last to re-enter the victim list. This realizes equal user experiences to all the users with the currently provided network halting GHG emission from new equipments.
This solution is suitable for the video downloads other than live streams (like downloading videos from youtube etc.). Keeping track of modifications to the priorities while creating victim lists will give more fairness to the users.
Figure 2. Prioritization Solution
4.2. Frame Deleting Solution
Another solution that would meet both requirements would be cutting video frames out of video streams without transcoding. Many of current video compression methods utilize prediction frames to display outputs to the screen. Generally, 3 types of frames are used: “Intra-coded picture” frame (I-frame), “Predictive picture” frame (P-frame), and “Bi-predictive picture” frame (B-frame). I-frames contain information of the whole picture, while P/B-frames contain differential information from the surrounding frames. The display outputs for P/B-frames are derived from I and other P frames. In other words, B-frames are not referenced by either I/P/B-frames for display outputs.
If we put nodes in the network that would only drop B-frames during network congestion, the frame rate of the video will drop. However, no information for the I/P frames will be lost, so there will be no noise on the video output. The frame deleting node only needs to know if the information sent is a B-frame or not, transcoding is not required and can be easily realized. Dropping B-frames to secure bandwidth may be one solution to give equal user experience to all users with little network expansion and environmental impact.
Figure 3. Frame Deleting Solution
5. Conclusions
Throughout this document, we have discussed the growth of video traffic in the future. If no solutions are proposed, the network expansion needed will give impact to the earth’s environment. To counter this, we have proposed two methods to lighten the network expansion. We propose that this document is accepted for deliverable 2 of Green Energy activity, “Study on potential enhancements of ICT.”
6. References
[1] Cisco Visual Networking Index, “Forecast and Methodology, 2009 - 2014” June 2, 2010,
http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/white_paper_
c11-481360.pdf >.
[2] Yasuhiro Yamada, Hitoshi Watanabe & Ming Ning Gu, “Main profile H.264 codec: A low power implementation for consumer applications” Oct 1, 2007,
http://www.eetimes.com/design/power-management-design/4013336/Main-profile-H-264-codec-A
-low-power-implementation-for-consumer-applications>.
- 1 -