• by Jeff Glueck , June 26, 2013

Anyone who has recently upgraded a smartphone from one that runs on a 3G network (say, the iPhone 4) to one that streams video and downloads songs on a 4G LTE network (for instance, the iPhone 5) knows anecdotally and intuitively that these activities are generally a better experience on the 4G LTE network. For consumers, 4G LTE is less prone to buffering and stuttering, and is one that, most of the time, just works. Wouldn’t it be nice if that were simply the end of the story, and that the problems that have arisen from deluge of high-bandwidth video traffic onto mobile networks just vanished once everyone migrated to LTE?

Unfortunately, while the situation is improving in the aggregate, this by no means allays the concerns of the many remaining frustrated video viewers who can’t get their YouTube or Vimeo clips to start. These consumers generally aren’t concerned about how crystal clear or “HD-like” video is once it finally arrives – they simply want it to arrive quickly and to stream without interruption. Today, consumers perceive their mobile video quality of experience (QoE) by transmission quality, rather than visual quality. They’ll reward an operator by staying with them, and even recommending them, if mobile video simply works.

When operators look to manage and optimize the over-the-top (OTT) video traveling on their networks, so that each individual consumer receives a first-rate quality of experience - even when they’re at the edge of a cell, inside a building or behind a wall - there are three approaches that can be taken:

1.   Operator-defined global optimization settings that are based on time of day or other static, outdated policies .

2.   Using “probes” in the radio access network (RAN) to identify congested cells – then focusing optimization only on those congested cells.

3.   Using the cloud to target network congestion on a per-user, per-stream basis, and optimize troublesome videos accordingly.

Option one is a legacy holdover from the recent past, in which expensive inline hardware was configured in mobile networks to reduce bandwidth on all videos at pre-defined times of day, effectively punishing all users on a given cell or even across the entire network whether there was congestion or not. It neglects to take into account the realities of modern mobile video traffic, where a previously uncongested cell may become overwhelmed due to an event or an unplanned spike in usage.

Option two, the RAN probe approach, is more surgical and can identify congestion where it is occurring on individual cells, yet is expensive to deploy and still suffers from a macro-level approach. Many quality of experience issues have nothing to do with cell-wide congestion, and are instead caused by impairments at the individual user level. This one-size-fits-all policy applies optimization across the board to everyone on a cell, while neglecting, say, a smaller number of individual users on a neighboring, uncongested cell who are abandoning their videos while cursing their operator.

It’s my belief, and that of many global mobile operators as well, that now that the technology exists to surgically optimize problematic video streams in the cloud on a real-time, per-user, per-stream basis, we’ve reached a new milestone in managing mobile video QoE. Individual users can now be saved from an unhappy video session, without bandwidth reduction for the other users surrounding them.

Operators also now have an insurance policy in hand that helps keep their users happy and satisfied with network performance, and not searching for reasons to blame their operator for poor video streaming sessions – as consumers are all too ready to do. Using the cloud to measure, quantify and mitigate video quality issues at the individual user level is more flexible, more cost-effective and delivers a much higher QoE than alternative methods. It’s enough to plug the remaining holes in LTE video quality, so that everyone -- you, me and that guy standing behind the wall  -- can stream video to their heart’s content.