Video streaming analytics has become indispensable to improve online video experiences in mobile operators.
The GSMA predicts that video streaming will account for roughly 75% of total mobile data traffic worldwide by 2023. As video formats have continued to evolve (with the emergence, for example, of 4K and 8K resolutions) so have end-users viewing habits and their demand for seamless quality, anywhere. Those are not, however, the only challenges that mobile operators will face in this ‘total streaming age’.
Measuring the Quality of Experience (QoE) obtained by their subscribers will be another one. With the majority of video content now being encrypted, traditional indicators for identifying QoE are no longer available. Mobile operators are practically left in the dark then, with limited visibility over service performance issues.
“While traffic encryption is necessary to safeguard individual privacies, it has the perverse effect of making it more difficult for mobile operators to derive critical buffering, resolution and stalling metrics.”
That, in turn, prevents them from isolating problematic locations, network elements and devices where most streaming issues are encountered. Taking proactive actions to optimise and distribute network resources more effectively becomes, thus, impossible.
One alternative to overcome those limitations, without compromising payload security, is to use machine learning or other pattern detection algorithms. They can predict key quality indicators per video session (i.e. initial buffering time, resolution, resolution changes, stall events, stall duration) based on visible extracted features from video streams.
Once that intelligence is developed, mobile operators must also deploy it into production. Doing so, on top of the voracious volume of video streaming data that they enable and without compromising the performance of their network analytics systems, is another daunting task.
Nevertheless, all that effort can pay-off. For mobile operators, conquering video analytics challenges means optimising network design and planning around key video streaming performance indicators. Ultimately, that leads to a systematic monitoring of overall video streaming quality which can promote best-in-class video viewing to an explicit element of their value proposition.
In this day and age, when mobile video streaming has become so ubiquitous, that is a major step forward in securing better subscriber experiences.