It isn’t just smartphone owners who are at the forefront of the mobile-video revolution, however. In 2011 the research firm In-Stat reported that 50 percent of tablet owners use their devices as their primary viewing medium for watching full-length movies and episodes of TV shows. And earlier this year Parks Associates stated that 60 percent of smartphone owners and 70 percent of tablet owners watch OTT video on their mobile devices at home, demonstrating the opportunity that is now available for savvy network providers. The home’s second screen is steadily becoming its first.

The solution

Traditional IP networks without an end-to-end QoS (quality of service) architecture provide best-effort service over a common, shared infrastructure, but any link or node in the network can experience congestion. The primary mitigation strategy is to drop packets proactively, which reliable networking protocols account for with built-in congestion-avoidance algorithms and retransmission mechanisms.

By default, complex wireless networks such as 3G have more places in the delivery chain that are capable of experiencing congestion, leading to increased latency and packet loss on mobile devices. In the case of UDP-based delivery (user datagram protocol) this issue manifests itself as severe client-side jitter, whereas with TCP-based delivery (transmission control protocol) it rears its ugly head as sender-side timeouts and retransmissions, and poor network conditions during media streaming can cause low network throughput, which in turn may cause the client’s media-input buffer to empty during playback (essentially, some segment of data doesn’t arrive on the client’s side in time for the video to be played back when the client wants it).

In other words, playback stalls out, waiting until enough content has been received for the buffer to be replenished; TCP-based protocols that don’t allow for quality levels to adapt to available throughput cause delays in playback, either at the beginning of a session or while it’s in progress. Therefore the details of these playback problems are the only factors that matter in determining how a mobile service provider’s network has impacted QoE. 

For adaptive streaming protocols delivered over TCP, network congestion will cause a client to request a lower bit-rate stream, which will result in a reduced amount of delays, or stalling, but will also reduce the visual fidelity of the video in question. And for unreliable UDP-based content, there are no retransmissions in the event of packet loss—dropped, corrupted and late packets basically vanish into thin air—resulting in many different quality-related issues, which can take the form of damaged data blocks, video stalling or jerkiness, a loss of synchronization between video and audio, or combinations thereof.

With so much for mobile service providers to consider when investigating the quality of experience that their subscribers receive, it’s no wonder that so many rely on tried-and-true metrics like network performance rather than actual subscriber QoE. Fortunately, a balance can be struck: advanced service providers around the globe are deploying OTT-video analytics that monitor, store, collate, and (of course) analyze data that pertains to their mobile-video network traffic. These analytics can now reveal the network impact of new devices and subscriber usage patterns, help plan and shape network expansion and inform optimization solutions.

To monitor mobile-video sessions and associated quality in real time over an entire network, Canadian service provider Telus worked with Avvasi to a system in place that supports all major OTT protocols, including progressive download, HTTP chunked download, real-time messaging protocol (RTMP), HTTP Live Streaming (or HLS, via Apple), and Silverlight Smooth Streaming (via Microsoft). The solution explores what Avvasi calls the delivery quality score, or DQS, and the presentation quality score, or PQS.

DQS reflects the impact of network delivery on QoE, abstracting the source’s quality so that real-time media sessions are inspected, monitored and reported. PQS does the opposite: it measures the quality level of a media session, taking into account device display and ignoring the network impact. Relevant key performance indicators (KPIs) covering audio, video and device are extracted from each media session, and Telus is currently finishing up plans for a multitiered service offering that includes video quality on user devices.