Monday, January 26, 2015

Broadband Internet everywhere, for cheap?

Broadband Internet everywhere, for cheap?  That's the vision of Elon Musk, CEO of SpaceX and Tesla.  A couple of weeks ago he announced his intent to create a $10 billion network of low earth orbit satellites as the first step in an interplanetary Internet backbone.  These satellites, presumably launched by SpaceX, would provide Internet access to every spot on the planet.  Late last week, he announced a $1 billion round of financing, led by investments from Google and Fidelity – the first step toward realizing that vision.

Current satellite-based Internet distribution is almost entirely done with satellites in geosynchronous orbit, about 26,000 miles above the earth's surface.  At that altitude a radio signal takes about 1/7 of a second to travel between the earth and the satellite, a delay that is the inevitable result of the speed of light.  That delay may seem small, but it adds up very quickly.  The simplest possible web page (with no images or any other non-textual content) requires a minimum of 5 earth-to-satellite-to-earth signal trips (for the geeks: 3 to establish a TCP connection, then one request/response pair).  Each of those trips takes 2/7th of a second, so 5 trips takes 1 3/7ths seconds – and that's the fastest you can get!  There are some tricksy things that can be done to reduce the number of round trips, but the absolute best one can do is still so many trips that browsing the web feels sluggish – a very different experience than one gets with terrestrial Internet connections.  The geosynchronous satellites do have one very compelling advantage, though: a single satellite can service an entire continent (North America, for example).

Musk's vision for satellites in low earth orbits would have quite a different operating characteristic.  First and foremost, they'd be less than about 1,200 miles high – I'm guessing 500 or 600 miles.  Below roughly 500 miles the atmospheric drag limits satellite lifetimes.  Higher satellites increase the speed of light latency, but also increase the “footprint” (the size of the area the satellite can service).  On the other hand, smaller footprints might be desirable, to limit the number of connections that a single satellite must be capable of.  So my guess is Musk will opt for the lower orbital altitudes, to optimize performance and minimize individual satellite size and complexity.  To guarantee global 24x7 coverage, Musk will need hundreds of satellites – even more if there is to be redundancy.  Hence the $10B price tag, even with assumptions about SpaceX drastically reducing the cost of a launch.

Suppose Musk actually pulls this off (and his track record is pretty darned good).  What would that mean for Internet access?  It would mean that no matter where you were – in a city or in Antarctica – you could get 100mbps+, low-latency Internet access.  Broadband for all.  At sufficient scale (not a small constraint, mind you), the $10B price tag will look like peanuts compared to terrestrial methods that require things like digging trenches and burying cables or fiber.  In other words, if Musk gets enough people on the system, it has the potential to be cheaper than existing broadband connections.  Much cheaper.  If the orbits are at 500 or 600 miles altitude, the latency will be very comparable to terrestrial networks – in other words, no performance downside.

Internet distribution could be the killer app for space.  Google and Fidelity are apparently persuaded that that's a good bet.

And Elon Musk will own the world :)

No comments:

Post a Comment