When Bigfoot Networks was acquired by Qualcomm towards the end of 2011, you could have been excused for thinking that Qualcomm was about to go into the game enablement business.
Instead Mike Cubbage, erstwhile CEO of Bigfoot, was at TV Connect this week with the job title of director of business development, networking business unit, Qualcomm Atheros and he was launching Streamboost, a Software as a Service phoenix that has emerged from the ashes of Bigfoot.
You know by the name that working at Qualcomm Atheros has mellowed the man somewhat, but the technology relies on the same ideas, even if it does not sing off the same hymn-sheet.
‘Streamboost started out as a form of QoS management that stopped gaming on a WiFi network being at the mercy of everything else. If you have 10 Mbps from your ISP to your home, and someone is watching a video, someone else is on a skype call and you are playing a first person shooter, you have no chance. You’re dead before you can move the cursor.’
And that’s more or less the problem that Bigfoot set out to solve and it solved it by creating a piece of software that can work with Atheros and some other WiFi gear, which will prioritize all the individual streams inside a home router. That way it can save those gaming sessions. ‘If you have a YouTube feed, it will just absorb more and more of your pipe. If your CDN sees you can operate at 10 Mbps, it will send more and more data, so it can get its job done faster. But if you tell the CDN clearly that you are letting it have 3 Mbps, then it will be happy with that, and not take more, or at last not much more.’
If you decide to watch a YouTube video it could be 360 pixels wide or 240 or 720 or 1080. The smallest might only require a few kbps and the largest might be happiest with 4.5 Mbps.
The real smarts in this product is that it acknowledges that it doesn’t have the power to use full Deep Packet Inspection to identify what type of activity is being used on your home network, so instead it looks for discreet characteristics that give it clues.
‘Our detection engine looks at packet attributes to detect which Apps are running. And that information will give you enough of a policy to better manage the data streams, so that users are not prevented from doing what they want to do with the internet. So we also have a policy engine running in the router as well. Our system takes 2 seconds of traffic to detect what type of data stream it is, and how many MHz of WiFi it needs.
‘The Policy engine determines how much bandwidth it needs to play optimally, and looks up that data from the cloud. It then reports back to the cloud how the performance worked out and we use that to update our policy engine. Effectively it stacks things in conduits,’ and Cubbage insists that it can put YouTube video, a torrent, 3 HD movies, skype, and a game all inside that 10 Mbps limit.
Cubbage said that it took two years and a couple of Maths PhDs to work out how to do all this and it has strong IPR and can work with most WiFi equipment. He then corrects himself as we ask about Broadcom, ‘No there is a problem with Broadcom,’ so we ask about Marvell, ‘and a problem with Marvell’ he says, ‘but pretty much everyone else.’
The use case for Qualcomm is much further reaching than simply protecting home gaming delays. On the internet of things and in machine to machine applications in the home, there will need to be many more streams than there are today, many of them smaller, but which need protecting. By laying down the principles now, and perfecting them with long term use, Qualcomm will be well placed to ensure that new home services such as wellbeing, health and security, are never starved of bandwidth by HD Youtube videos.