Top Banner

Click here to load reader

Disruptive Analysis White Paper - Consent-Based Video Optimisation

Oct 15, 2014



White Paper commissioned by Yospace on the role of user and publisher "consent" in mobile video optimisation. Mobile operators should not deploy solutions that modify traffic without the direct permission of user and/or content publisher - that is certainly not "optimal" in any sense. Instead they should work with new techniques such as CDNs and adaptive bitrate streaming to improve user experience & manage congestion on the network

Consent-based video optimisationEnabling a new role for mobile operators in the video value chainA Disruptive Analysis thought-leadership paperCommissioned by Yospace

February 2012

Author: Dean Bubley Contact: [email protected]

Disruptive Analysis Ltd, February 2012

Consent-Based Mobile Video Optimisation


Disruptive AnalysisIntroduction

Dont Assume

For mobile operators, video is both a threat and an opportunity. Its use drives the purchase of data-plans by smartphone and tablet owners, but its growing popularity threatens to clog up cellular networks with extra traffic. It is possible to reduce the impact of Internet video fairly quickly, by using techniques such as optimisation to compress video streams in the network, or pace their delivery, which reduces some traffic load. But this can alienate content companies, which dont like to see their material being altered - especially by operators who then try to sell them enhanced QoS. It can also create issues where customers have paid for HD-quality content, which the network downgrades without permission or notification. In addition, few implementations only squeeze down video just when there is actual network congestion, which could be seen as reasonable by regulators most attempt to optimise across-the-board. Partly linked to this, content publishers and aggregators are increasingly now using applications or new video formats which self-optimise to counter network congestion or unavailability. Techniques such as adaptive bitrate streaming are becoming popular, as are various forms of encryption and DRM (Digital Rights Management). These approaches may well also have the side-effect of bypassing the optimisation boxes or conflicting with them. The net effect is antagonism between stakeholders, rather than cooperation. At the same time as this so called OTT (over-the-top) story is playing out, some operators are thinking once more about developing their own video offerings. Despite indifferent success with mobile TV in the past, there are some signs of renewed interest in developing channels or video-based apps to provide content both to their own subscribers, and possibly to mobile users on other networks. Certain of the constraints that stopped mobile TV from gaining success in the past have been removed, and different use-cases and are taking shape. If operators can make their new video offerings inherently more network-friendly than other sources, then they can develop new business models and revenue sources while also providing their mobile data customers with an all-round better mobile video experience. Depending on the operator, it may also form the basis of future partnerships with content publishers and advertisers to create more integrated and interactive experiences, or paid-for enhanced delivery. Done properly, it should also be able to cope with the broadening universe of new smartphones and device types, while requiring minimal intervention to existing billing or network policy infrastructure. This paper examines some of the key trends in mobile video the tensions between the players, the realities of new video formats and delivery mechanisms, and the shifting roles of network operators and the traffic-management strategies required to run their mobile broadband networks. Note: This document has been commissioned by Yospace, but represents the independent & consistent viewpoint of Disruptive Analysis (as at time of publication). Disruptive Analysis has retained full editorial control over the content and stance. Disruptive Analysis Ltd, February 2012 Consent-Based Mobile Video Optimisation 2

Disruptive AnalysisNetwork congestion & pressure points

Dont Assume

The rising traffic levels on 3G and 4G networks are well-documented. Mobile operators frequently report high rates of growth in data volumes, while studies and forecasts from network vendors such as Bytemobile, Ericsson and Cisco highlight the role of video content and forecast huge future growth. While some of this data can be challenged (especially the predictions, given recent introductions of capped dataplans), there is nevertheless a real problem in some networks, and an imminent issue in others. That said, there is a broad diversity of both causes of network congestion and its potential mitigation. Some issues come as a result of signalling traffic from short data bursts or connections, while others stem from large Internet video downloads. Some problems relate to capacity limits in the radio network, others are driven by backhaul or core network constraints. And despite the efforts of many network vendors trying to sell extra intelligence such as application-detection and throttling/QoS measures, most of the real impact on lowering network load has come from billing and charging. Encouraging end-users to change their behaviour for economic reasons, rather than using network smarts to do it for them, seems to have been effective. Part of this reflects the difficulty of really understanding applications in the network: categorisations of data into broad buckets such as video, web and P2P are often arbitrary, overlooking mashups and combinations of services. Video is not an application per se but a component of many. A mobile web page may include a video-based content clip or advertisement, a web conferencing system might include telepresence, and an aircraft maintenance application might incorporate video-based instructions and examples. Given the strategic importance of mobile data, it is unsurprising that numerous vendors, across the value chain, have touted approaches to fixing the problem of traffic management, often with video as a specific focal point. There is no one size fits all method for treating mobile video traffic each of these has its own specific requirements. Some will be encrypted, some will be tagged to prevent caching, some will be streamed and some will use adaptive bitrate approaches. Increasingly, strategic combinations of these approaches will be used, rather than individual tactics. So it is important that solutions fix the right problem, in the right way and ideally also support current and incremental revenues for the operator.

Non-consensual optimisationAt the moment, many mobile operators around the world have implemented so-called transparent content and video optimisation. Typically, this involves a dedicated box sitting between the operators core network and the Internet itself on what is called the Gi interface. Sometimes, the optimiser is just a function of operators gateway node itself (the GGSN). In this scenario, the optimisation acts as a proxy between the user and the original content source on the Internet. It takes in content (in this instance video) on the northbound side, processes it, and then sends it on to the end user. The user is

Disruptive Analysis Ltd, February 2012

Consent-Based Mobile Video Optimisation


Disruptive Analysis

Dont Assume

typically unaware that he or she is seeing the optimised content rather than the source itself. There are various functions that mobile video optimisation boxes can perform. They can transcode data, for example from one codec to another. They can compress it, in various ways (eg dropping resolution, or frame-rate), delay it, or potentially even modify it for example by inserting advertisements. Each vendor has its own specific use cases and subtleties, but the general story is similar: (Realistically) attempt to reduce network congestion by sending less video traffic to the user than might otherwise have occurred (Potentially) reduce costs of upgrade or opex needed for the radio and/or backhaul network (Hopefully) improve user experience by reducing or mitigating the incidences of stalling (Optimistically) offer the operator a chance to monetise (i.e. charge) the OTT video content providers for sending streams over the network

It is important to point out that this form of optimisation does indeed reduce the overall amount of traffic on the network, so it is by that definition successful. However, the actual effects on perceived quality are relatively unpredictable especially to the original content provider. The user may end up watching a lowerquality stream so that buffering pauses are reduced but the video publisher does not know this, and neither do its advertisers. The user may have actually paid for an HD stream, and been prepared to wait for it to download into the buffer before watching it yet the network may have assumed it was being viewed in real-time and unnecessarily reduced the quality. Disruptive Analysis believes that this form of non-consensual optimisation has serious flaws. Ideally, both the user and the content publisher will have agreed to what is being done to the data flow (after all, the user is paying for it), and they will also be able to see what impact that optimisation has on the experience. It is interesting to note that various of the optimisation vendors are now evolving their approach, to differentiate amongst themselves and add value to their boxes, and to better fit with the realities of the video and content world. Many have developed additional use cases for their kit, such as attempting to measure and control video stalling by observing the behaviour of TCP/IP connections. Others vendors have been discussing ways to monetise third-party video content charging video publishers for enhanced QoE or prioritisation, for example. As yet, such notions have seemed superficially attractive, but it is not clear that the business-model promises match up to the practicalities of the technology.

Searching for new video business modelsSo, it is undoubtedly true that most of the volume traffic on mobile 3G/4G networks is video-based, although on some networks the bulk of it is