O3b Networks LEO Satellite Internet

[wjbarnett]

Do you see any possibility of LEO-based ATC satphone/cellphone combos making much headway into the market?

I've not studied LEO-based systems. But my first thought on the likely issues are a) (costs for) enough sats to provide continuous coverage with sufficient # of spot-beams, but small beam-size, and b) spectrum allocations (can't conflict with GSO sats, in any orbital position). I believe (but not fully current on latest FCC rule making) all ATC spectrum is currently GSO compatible. Plus c) the usual issues faced by every global player (ie highly variable regulatory environments). The uplink power reduction is a benefit for LEO-based sats.

Jon, My knowledge base is traditional cellphone & broadband data networks, and business models. But when a good client wanted a Sat/ATC tradestudy, I jumped in (with a comsat knowledgeable partner) and learned a lot. But that project completed 4 years ago with the client declining an investment offer, at least partial due to our study results.

[MikeAtkinson]

I took part in the WISDOM (Wideband Satellite Demonstrator of Multimedia Services) EU funded project in the late 90's.

We looked at GEO, MEO and LEO constellations. This was both about demonstrating end to end connectivity, and doing system design work. We built two ground stations. I did the bandwidth allocation algorithms and wrote both a simulator of one beam (thousands of users with up to 1000 active at any one time). It could also connect to the ground stations, controlling one as the network side and one as the user. Unfortunately the breadboard satellite system was late and we did not get to try out the entire system end-to-end.

While we had great confidence that the system was technically sound, I and many others on the project had grave misgivings about its commercial potential.

We looked really hard for a solution that would allow GEO satellites to be used first and then replaced/supplemented with MEO and then replaced/supplemented with LEO satellites. We could not find a solution which kept the ground station cost reasonable.

The problems we found included:

The service would cost a reasonable amount, probably more than broadband wireless or cable networks in towns or cities. So anywhere where there was a moderate density of (reasonably wealthy) population the satellite network would loose out to terrestrial solutions.

The system worked best when each beam had just the right number of subscribers, too few and the revenue would be low, too high and they would contend for bandwidth and get a degraded service.

Within Europe (this was a EU funded study) the potential user base would be continually squeezed by extensions of terrestrial networks. Realistically we could expect only a few 100,000 users at best, those in the most remote locations. This wasn't enough to pay for the system, even with EU subsidies for universal access to services.

For LEO and MEO systems most of their time would be spent over the ocean, where there would be very few subscribers.

Over much of the world we had the problem that most of the population were too poor to afford access, most of those wealthy enough live in cities where they would have terrestrial broadband. We would be looking at targeting the rural rich - a small, hard to reach segment of the total broadband market.

It would require marketing to almost every country at the same time, building up a user base very quickly.

When the required user base had been reached in an area there was not the flexibility to increase it without degrading the service for everyone.

A partial service would not be very good, so selling to users could not realistically take place until the constellation was in place (for LEO and MEO). What happens if a satellite fails, how quickly can the gap in coverage be filled. What would that do to the user base.

It is over a decade now since I worked on it so the details are fuzzy, however IIRC we reckoned that obtaining >20% capacity factor would be a major achievement.

These (and probably other risks I've forgotten), indicated that the risk of a financial failure were quite large, the potential gains would need to be equally large, even with optimistic assumptions this seemed unlikely.

In the end the company I worked for (Nortel) pulled out, we reckoned that if there was a broadband satellite system, we could bid for ground segment business just as well even if we were not a consortium member, and bid for the ground segment of rival systems as well.

[Robotbeat]

...
It would require marketing to almost every country at the same time, building up a user base very quickly.
...

Yes, this is a good point. I think the most likely scenario for an uber-sat-phone network would be an existing cellphone provider who already has fingers in every single market in the world, and could even distribute compatible phones well in advance to switching on the satellites. It would take much more capable LEO satellites than we currently have. That means much larger antenna arrays (enables more efficient spatial multiplexing to enable many more users per satellite than currently possible... each Iridium satellite is capable of 1100 concurrent active users) and more power, among other things. In very high population density areas, it'll still make sense to have cell towers.

[A_M_Swallow]

{snip}

Within Europe (this was a EU funded study) the potential user base would be continually squeezed by extensions of terrestrial networks. Realistically we could expect only a few 100,000 users at best, those in the most remote locations. This wasn't enough to pay for the system, even with EU subsidies for universal access to services.

You are assuming that subscribers are people, they could be towns and big companies.  The satellite would then be acting as a trunk network.

From this page
http://en.wikipedia.org/wiki/Earth_Population
The current world population is estimated to be 6,885,100,000, round up to 7b.  If you design for 100,000 ground stations each ground station would supply 7,000,000,000 / 100,000 = 70,000 people

There are plenty of towns that size.  Cities could be divided up into districts.  There may be sufficient ground stations left over to have individual ships and very isolated buildings as subscribers.

[wjbarnett]

A broadband link shared by 70k people is pretty low bandwidth per person/device (especially in today's world of smartphones!). And who would guarantee that they would sign up to pay for this system? And why, since the existing terrestrial broadband systems are already serving the vast majority of these people that in towns or cities?

[Robotbeat]

A broadband link shared by 70k people is pretty low bandwidth per person/device (especially in today's world of smartphones!). And who would guarantee that they would sign up to pay for this system? And why, since the existing terrestrial broadband systems are already serving the vast majority of these people that in towns or cities?

Those 70,000 people aren't going to be fully streaming all the time. All consumer (and many non-consumer) internet connections are over-subscribed, especially the wireless ones. Say you sell the connections as being 700kbps, each person may only get 15kbps constant throughput, but since everyone isn't always using it, you get a lot more throughput, and the provider still only has to do ~1Gbps per satellite, which is more than reasonable for a next-gen system. Trust me, this is already how it's done with smartphones, hence the whole "5GB per month" thing for many existing smartphones (which works out to about ~15kbps on average, interestingly) with $30 per month plans.

This still goes one with wired connections, too, but usually isn't as bad. But terrestrial wireless data connections that are $30/month for 5GB/month are the competition. I think it's possible to beat that.

(and actually, most people don't come near the 5GB/month for their 3G devices... not saying they won't or that it wouldn't be nice not to have greater limits, but that's the reality right now)

[wjbarnett]

Trust me. It's not average traffic that matters, but rather peak traffic. Traditionally we used "busy hour" but these days it about "peak minute" speeds (that is a sum of rates, NOT volume) that drives the sizing of broadband networks.

[Robotbeat]

Trust me. It's not average traffic that matters, but rather peak traffic. Traditionally we used "busy hour" but these days it about "peak minute" speeds (that is a sum of rates, NOT volume) that drives the sizing of broadband networks.

I believe you!

[Danderman]

Scads of technical information in this FCC application.

[zaitcev]

So... Are they dead like LightSquared or not yet?

[Skyrocket]

So... Are they dead like LightSquared or not yet?

As two launches with each 4 satellites are scheduled for this year, they seem to be alive. (http://forum.nasaspaceflight.com/index.php?topic=6114.msg994277#msg994277)

May 29 - O3b (4 sats) - Soyuz-STB/Fregat-MT (VS05)

Late July-Early August - O3b (4 sats) - Soyuz-STB/Fregat-MT (VS07)

[Zond]

O3b's First Satellite, Fully Tested and Ready for Launch!

[zaitcev]

First batch was launched today, see launch thread:
 http://forum.nasaspaceflight.com/index.php?topic=31727.105

And I almost wrote them off for dead at the start of the year! People work, not post press releases.

[sb]

There's a good article on theregister.co.uk on this and some interesting stuff in the comments from someone who was involved in the design of the ground stations (I think, read it a couple of days ago). Will add link when on a PC

[R7]

Here is The Register article.

Ground station needs two 3m tracking dishes for uninterrupted service, preferably third as backup. Estimated cost several million $ (are they really that expensive??).

Hoping these guys make it and congrats for successful launch!