Video already constitutes more than 50% of mobile traffic and continues to grow. Rocket Optimizer cures buffering delays by intelligently dropping video quality.
With limited radio spectrum and the rapid growth in data consumption, wireless carriers are having trouble keeping up with data demand on their 3G and 4G networks--even after investing tens of billions a year in upgrades. The transition to smartphones has brought the full Web experience to mobile phone Web browsers, and that entails a lot of video streaming, despite the fact that mobile networks aren't suited for video on demand. The resulting congestion often causes delays in buffering when users are trying to watch streaming video.
Skyfire offers an alternative to buffering delays with a carrier solution called Rocket Optimizer. It is a cloud-based service that buffers and dynamically converts streaming video to a lower bit rate with an average size reduction of 50% to 60%. The concept of additional compression for Internet Web content is nothing new and has been used as far back as the AOL dial-up days when images had to be compressed further to run well on 0.028 Mbps dial-up modems. Though images no longer constitute the bulk of the network traffic, video already accounts for more than half of wireless network traffic.
Given the sometimes contentious nature of network management, especially when content is being reduced in quality, I interviewed Skyfire CEO Jeff Glueck to better understand how the product works. I was pleased to hear that the additional compression is only applied when there is network congestion. More importantly, this prevents video streams from being slowed by congestion control mechanisms within the network. So instead of buffering pauses in the video stream, the video bit rate is temporarily reduced in size.
Taking traffic data from the mobile networks as input, Rocket Controller might choose to route video data through the Rocket Optimizer to keep traffic flowing smoothly and fairly to app viewers.
Furthermore, the additional compression is applied fairly between customers and content sites. If everyone is trying to view the same bit-rate video and the network's capacity can't support all of them concurrently, then everyone's bit rate is shrunken enough to make all the streams fit. If 10 customers on a cell tower are viewing lower bit-rate videos while one customer views an HD video stream, the HD video stream will undergo additional compression while the others are left alone. It wouldn't be fair to apply additional compression to video streams that are already low in bit rate.
From a usability standpoint, buffering delays are far more disruptive than a small loss of video fidelity. Though many content sites offer a selection of video bit rates and resolutions, granularity is often lacking. Netflix, for example, might abruptly drop an HD stream from 2 Mbps to 0.5 Mbps when the network is congested, even though it might support 1.5 Mbps. A dynamic transcoder can in some cases offer a higher quality of video to the mobile device when an in-between bit rate with more optimal resolution is provided.
Skyfire shrinks video to the optimal resolution of the particular client device. Many first-generation smartphones such as the iPhone 3GS are capable of displaying only HVGA (480 pixels by 320 pixels) which means widescreen content is rendered at a 480-pixel-by-270-pixel resolution. The benefit of a higher resolution and higher bit rate video stream at these lower resolutions is less noticeable and a dynamic compression scheme seems to be a reasonable compromise.
On a related note, Skyfire also offers a solution for consumers that solves the Adobe Flash compatibility problem on iPhones and iPads. The Skyfire iPhone and iPad apps tap into Skyfire's transcoding cloud, which can convert flash videos to an iOS-compatible format.
While Skyfire's carrier solution is already fairly reasonable, I asked Jeff Glueck if users could have a way to opt out of the system. For example, some wireless customers might be willing to let a higher bit rate and higher quality video buffer while they do something else rather than take a lower bit rate in real-time. The vast majority of customers will likely prefer to avoid buffering delays, but an opt-out capability would make everyone happy. Glueck responded that this might be a good feature and that it is probably a good idea for carriers to implement this, but for now users who want maximum fidelity and minimal delays are better off using a Wi-fi connection linked to a wired Internet service.