Is Loon flying?

December 3, 2014 11:50 AM

Google's balloon-based internet seems to be working. After some hiccups, one bad demo, and lots of redesign, Google's Project Loon (previously) is bringing some internet to some people in the developing world.
Fluffier socks play a crucial role. (SLS)
posted by doctornemo (25 comments total)

I've sometimes wondered if you could generate fresh H2 by solar electrolysis from the moisture in the air to keep a balloon hovering indefinitely. I even once posted this idea to HalfBakery.
posted by RobotVoodooPower at 12:07 PM on December 3, 2014

I wonder how fast the connection is. If balloons can solve the 'last mile' problem, they should start launching them over the continental US.

I wonder if that would be cheaper than the cable company route (buying off state legislatures).
posted by leotrotsky at 12:18 PM on December 3, 2014 [2 favorites]

The balloons are designed by Raven Aerostar, who also offer persistent surveillance aerostats and closely partner with the US government.

The Snowden docs have revealed Google's close ties to the NSA.

Anyhow, red flags - I'd be strongly inclined to down these if I were leading a "developing" country.
posted by ryanshepard at 12:20 PM on December 3, 2014

ryanshepard: They could be sitting on a throne made of the skulls of fallen angels; they'd still be better than Comcast.
posted by leotrotsky at 12:24 PM on December 3, 2014 [16 favorites]

I'm still just massively skeptical of this whole endeavor as anything other than a very niche application.
posted by BlackLeotardFront at 12:26 PM on December 3, 2014

If I were a Fox News script writerjournalist, I'd be working hard to flip this into some sort of Google giving free internet to the terrorists!!! outrage.
posted by Thorzdad at 12:32 PM on December 3, 2014 [1 favorite]

This project represents the best of GoogleX. It's a ridiculous idea on the face of it, enormously complicated and implausible engineering with a giant set of regulatory and business challenges. But it's also really cool and solves an important problem in an interesting way. I'm glad Google is using some tiny bit of its giant mountains of cash to try this out and I'm impressed they have had some positive results.

Wikipedia has a list of GoogleX projects. They are: Google's self-driving car, Project Wing a drone delivery project, Google Glass eyewear that includes a screen and camera, Google contact lenses that monitor glucose in tears, Project Loon which provides internet service via balloons in the stratosphere, An airborne wind power company called Makani Power, Lift Labs makers of a tremor-cancelling spoon for Parkinson's patients, An artificial neural network for speech recognition and computer vision, The web of things.
posted by Nelson at 12:40 PM on December 3, 2014 [6 favorites]

But it's also really cool and solves an important problem in an interesting way.

The important problem of there remains some people left on this planet that Google hasn't indexed. quantified, and shoved ads at yet, presumably.
posted by entropicamericana at 12:54 PM on December 3, 2014 [3 favorites]

Well said, user 58964!
posted by shponglespore at 1:11 PM on December 3, 2014

The important problem of there remains some people left on this planet that Google hasn't indexed. quantified, and shoved ads at yet, presumably.

You tell'em, ace.
posted by kbanas at 1:12 PM on December 3, 2014 [2 favorites]

Anyhow, red flags - I'd be strongly inclined to down these if I were leading a "developing" country.

I understand that the NSA uses phones and computers to spy on people, which is why in my third world country, I will ban all electronics and electricity. It's a bit of a bother, but at least our privacy will be preserved.
posted by happyroach at 1:31 PM on December 3, 2014 [6 favorites]

entropicamerican, you left out the fortune Google's going to make selling tin foil hats once they start beaming all that deadly radiation from their death balloons.

Yes, if this technology turns in to a commercial product it will no doubt come with a host of business and regulatory problems. My guess is it will be a lot like cell service, although maybe in the hands of someone more imaginative and less predatory than Verizon and AT&T. But that's many years away. I'm most excited by the audacity of a research project like this at a for-profit company. I suspect like self-driving cars it's 10–20 years away from being something they could actually ship.
posted by Nelson at 1:43 PM on December 3, 2014 [3 favorites]

If there's one thing this world needs, it's more automobiles.
posted by entropicamericana at 1:51 PM on December 3, 2014

If there's one thing this world needs, it's more automobiles.
If there's one thing this world needs, it's SAFER automobiles
Fixed that for you.

I'm completely with you about the need to cut down on emissions, but I don't think that a goal of having fewer cars on the road needs to be at odds with self-driving cars. If they work as advertised, self-driving cars have the potential to save tens of thousands of lives. Will this mean more cars on the road? Maybe. But the likely positive benefits of self-driving cars outweigh the drawbacks considerably.
posted by Green Winnebago at 2:03 PM on December 3, 2014

The article claims the balloons would be cheaper than setting up a network of cell towers, but I'm mostly curious how it compares to satellite internet. Is it competitive in terms of setup and maintenance costs (for providers and end users), bandwidth and connectivity, and overall accessibility?

The project is framed as a way to get internet access to people in remote areas. I also wonder if it's a feasible replacement when existing internet access gets shut down. I imagine state actors wouldn't have much trouble taking down these balloons, but maybe they'd be a handy temporary solution during or after a natural disaster, or to circumvent localized mobile phone jamming during a protest.
posted by twirlip at 2:03 PM on December 3, 2014

twirlip, satellite isn't always an option, depending on location and conditions. The base stations are also very expensive to set up and maintain (I believe) - it's good for some places but not others. I wrote an article a while back (mefi's thewalrus was very helpful) that looked at some of this stuff - you might find it interesting.
posted by BlackLeotardFront at 3:32 PM on December 3, 2014 [1 favorite]

Current satellite Internet is really terrible. Hughes, the main US satellite provider, works on geosynchronous satellites. Those are way up at 35,000km or about 240ms round trip time. Not only does that latency make real time interactions impossible it also screws up TCP bandwidth because of some oddities in TCP/IP design. (Even worse, usually satellite internet requires a wired phone line for the return channel.)

I'm not sure why we don't see satellite Internet with a constellation of lower satellites. More expensive and more difficult since the satellites are moving. Loon is somewhere between a cell tower and a satellite. The real challenge is making it low cost and manageable and reliable.

Also it's not just "deepest Africa" deployments; I'd love to have this technology in California. I have a place in Grass Valley that has terrible fixed wireless Internet, my only option. Despite living less than a mile away from a brand new big fat fiber optic backbone. AT&T and Comcast refuse to extend service to even slightly rural areas, and local wired competition is hard to come by.
posted by Nelson at 3:48 PM on December 3, 2014 [1 favorite]

I can pretty well guarantee you that, after the initial R&D costs are expended, running this balloon program would be less expensive than setting up enough cell towers to cover the same area. Average cost to install a single cell tower: $250,000, sometimes more. That's not including the rent paid to the landowner (it varies and is negotiated with each landowner but is typically more than a car loan payment, but to be paid basically in perpetuity), and the electricity the towers use (those 30-foot long equipment sheds at the base of the tower include so much electronics that they also have to run an air conditioner just to keep them from overheating). There also is regular maintenance of the towers and associated equipment, requiring technicians to drive out to all of them, etc. I think solar powered balloons would just have to be significantly cheaper than all that, at least on the scale being proposed by Google.

(Disclosure: For many years, I did environmental legal work for one of the big cell providers in the U.S. That's where I learned the above.)
posted by JimInLoganSquare at 3:56 PM on December 3, 2014 [3 favorites]

This is really cool. The fluffy socks part, especially. Still, this bit exemplifies Google’s typically chicken-shit approach to any problem that's not engineering driven: The company is famously far more interested in solving complex engineering problems than it is in dealing directly with consumers or lobbying bureaucratic agencies. Google has a blind spot for predicting how their HR projects affect populations outside their comfort zone.
posted by migurski at 4:47 PM on December 3, 2014

Nelson, the biggest problem with the vast majority of commercial satellite internet is that they're using C/Ku-band gear that requires big-ish, dedicated dish antennas pointed at satellites in geo orbit- you're looking at 600-700ms total latency when you consider going out there and back to wherever your ground station is.

There are some startups working on lower-flying multi-satellite constellations like you mention- O3b is already is up and running, I think? The downside to these multi-satellite systems is that they require a much more complex antenna compared to your average C or Ku-band setup, but it's a huge improvement bandwidth and latency wise.
posted by AaronRaphael at 5:10 PM on December 3, 2014 [2 favorites]

McCaw, Bill Gates, and Boeing were working to create a low-Earth tiny-satellite internet company called Teledesic, back in the 1990s. I knew someone who was working on the satellites. It didn't succeed, unfortunately.

Elon Musk wants to try this idea again.

There's already a low Earth orbit satellite company for phone calls, Iridium, and their next generation satellites will be able to provide 8Mb/s anywhere in the world, but they are an expensive high-end service, which won't be much use to developing nations.

You can't get much cheaper than a balloon.
posted by eye of newt at 10:19 PM on December 3, 2014 [1 favorite]

Nelson: "I suspect like self-driving cars it's 10–20 years away from being something they could actually ship."

Yeah, Google's so-called "self-driving cars" are an interesting analogy here, since they're really no such thing. Google still hasn't managed to make a single test car drive autonomously; they've just poured money into hyper detailed maps of an area around their headquarters and pointed out the obstacles to robots just complex enough to avoid things that have been pointed out to them beforehand. The cars themselves are not really impressive at all, since there's no chance whatsoever that a similar car will ever be able to drive in any area that hasn't been meticulously mapped at great expense. I suspect the "self-driving car" project has two goals: first, it's aimed at providing a nice touchy-feely progressive-seeming future dream as advertisement for Google as a company; second, under the hood it's actually a research project into what costs and applications hyperspecific mapping might have. Either way, it seems extraordinarily unlikely that Google will ever manage to create a car that actually drives itself. The obstacles are too great, and they apparently haven't really tried to solve any of them in a real way.

So I tend to wonder about Google X projects. They don't seem to have a tendency to be earnest attempts to introduce technology. Something else is always going on behind the scenes.
posted by koeselitz at 2:50 AM on December 4, 2014

The Snowden docs have revealed Google's close ties to the NSA.

So, full disclosure, I'm a Google employee. I don't work in anything security related, and maybe I'm subject to a little organizational bias. But I keep hearing people talk about close ties between the NSA and Google, and I just don't see it.

For one, if it was that close a working relationship, they would have sent someone actually in charge of something instead of Sergei. Google has been under pretty consistent, concerted attack from foreign actors for years now, so it honestly seems reasonable they would need to work with the NSA to identify and handle threats. There's just no other federal-level agency with that kind of expertise, and this isn't something the private industry should be handling alone. I guess in my ideal world, the NSA would be strictly an organization that helps American businesses properly safeguard their IT infrastructure instead of what they are now - a kind of thinly veiled Stasi reference from a William Gibson novel.

The Snowden docs shocked Google engineers, and kicked off a massive redesign of Google's security architecture. Five years ago, it really was reasonable to assume that private leased fiber optic lines were secure. Thank god for Snowden, because now we have proof that's no longer true.
posted by heathkit at 2:01 PM on December 4, 2014

BlackLeotardFront, thanks for that link. One reason I'm curious about this is that my employer works with some libraries in interior/northern BC where satellite is the only real option, and poor connectivity is a recurring issue. We had to pass up working with some folks in Nunavut for the same reason. For those of us in major urban centres, it can be easy to forget how far we are even in North America from universal decent high-speed internet access. I'd love to see serious public investment in building out proper broadband infrastructure in remote areas; I think you're right about the need and about the barriers to making that happen. It would be pretty cool if something as seemingly farfetched as balloon-based wifi could fill the gap in the meantime.
posted by twirlip at 2:21 PM on December 4, 2014

I'm still fairly chipper about constellations of LEO/MEO satellites. There's a lot going on with metamaterial antennas, which promise very high gain from electrically small devices plus the ability to steer beams from non-moving, mass-produced devices. Plus there's been a ton of work on 5G-ish spatial/temporal multiplexing, with both that and the antenna stuff well down the lab-to-shelf pipeline, from what I can see, and LTE has helped promote increases in efficiency and cost reduction with the sort of RF power engineering that gets more important as you push markets into new, commercially and geographically challenging places.

Further along, optical inter-satellite links should mean you can get a lot more bandwidth routed around the atmosphere than you can through, over long distances.

Some of this also benefits Loon-style and ;loitering drone relays, as well as terrestrial towers, but I think historically the race goes not to the 'really good for difficult cases' solution but the 'works really well for lots of people and quite well for the rest' and if you can get decent amounts of bandwidth from LEO that's going to cover the planet by default. It worked the other way for Iridium, which got the lumpy end of the stick from GSM's invention of a global terrestrial handset, infrastructure and services market.

In the end, I've no doubt that a mixture of delivery mechanisms will prevail, as it does now. Developments in how they all glue together may be a defining factor in what wins - roaming per se may not have been the killer for GSM, but having a global standard necessary to make that work gave it the economy of scale that did.
posted by Devonian at 4:38 PM on December 4, 2014 [1 favorite]