Dear Sir,

I read your interesting paper on satellite internet access/backhauling costs.
I also read the announcement of Other 3 Billion Networks and their MEO benefits and characteristics which are completely different to the ones you consider in your paper. Is this due to a completely new generation of satellite technology?

Could you please comment on the following most striking features:

A) 1Mb/s to 10Gb/s per backhauling connection link.

B) They state the system made up of 16 satellites for the initial phase to support 2133 transponders, this would mean 133 transponders per satellite, in your case you state 15 transponders per satellite.

C) They also say that the cost of internet backhauling would be 500$ per Mb/s per month against 4000$ of competing systems.

Thanks

Iaio

For those that would like to read more about the O3B project go to their web site here: https://www.ses.com/o3b-mpower/o3b-mpower-technology

8 satellites are launched simultaneously and then deployed in pairs from the upper stage. Only two launches will therefore be needed to put up the 16 satellites. There are 12 spot beam antennas shown on each satellite, presumably each actively steering.

The satellites are all in circular orbit, above the equator, height 8000 km, so will be seen to be moving forwards, eastwards. Ground users will have to keep switching backwards to more westerly satellites every 15 minutes or so as the satellites march eastwards. This page https://www.satsig.net/O3b/O3b-orbit.htm should give you some idea.


Quote:

A) 1Mb/s to 10Gb/s per backhauling connection link.

Assuming use of 500 MHz in each of the bands 19.7-20.2 and 29.5-30.0 GHz and single polarisation, each spot beam could have 500 MHz bandwidth. If 300 MHz is used for downloads, that could provide 936 Mbit/s using 16APSK DVB-S2. Times that by 12 beams and you get around 10 Gbit/s which seems reasonable. It might be much more if they use more of the Ka frequency band or dual polarisation. DVB-S2 has the advantage it can drop back to lower modulation/coding during rain or to cope with beam edge or customer earth station deficiencies.


Quote:

B) They state the system made up of 16 satellites for the initial phase to support 2133 transponders, this would mean 133 transponders per satellite, in your case you state 15 transponders per satellite.

Strange number that 2133. My guess is 1 downlink and I uplink transponder in each of the 10 customer beams.

I believe that two satellite antenna will be used for gateway feeder links and that 10 satellite antennas will be used for 10 customer sites, each covered by a spot beam.

Will it work ? Much depends in the customer antennas, tracking (1.2m - 7.6m) and working in pairs. The satellite power (watts) per transponder is also important and again I can find no information.


Quote:

C) They also say that the cost of internet backhauling would be 500$ per Mb/s per month against 4000$ of competing systems.

I am pleased to see someone trying to get the costs down. The present prices of $4000 to $7000 per month per 1 Mbit/s are too high for most people. Much will depend on how cheap they can get the satellites manufactured and launched and the financial viability will depend on how quickly they can build up the number of customers paying. They will be watching the progress of Wildblue, Ka-Sat and ViaSat-1

I hope this O3B project does not go the way of other LEO and MEO satellite projects of the 1990's.

If anyone knows more about the proposed O3B customer antenna, how the satellite antennas will be pointed or the ground infrastructure please say.

Best wishes to all involved.

Best regards, Eric.

From their website:

"Each transponder carries up to 216Mhz forward and 216Mhz in return allowing for over 600Mbps (~OC-12) of IP trunking."

This translates into ~2.778 bps/Hz which looks more like a 8PSK 0.95 TPC with 2.85 bps/Hz.

That is interesting.

I was thinking in terms of quite small customer antennas and a 216 MHz wide transponder would need very high watts power on the satellite downlink. Does anyone make modems that receive at around 200 Msps with 8-PSK?. I imagined several narrower transponders, each at saturation and with moderate wattage power, and also with bandwidths corresponding to feasible symbol rate for cheap modems. I can't work out how they get 2133 transponder in total (16 satellites, each with 12 beams).

If they have 1 transponder per 216 MHz I wonder where they are getting the frequency range from to put 2133 transponders in orbit. Maybe they are using more bandwidth like 2.5 GHz up (25-27.5 GHz) and 1 GHz total down (18.3-18.8 plus 19.7-20.2 GHz). They could fill 1 GHz on the downlink with 8 transponders, each 250 MHz wide and dual polarisation.

It has set me thinking about another matter - interbeam interference. 12 beams from the same satellite will require some clever interleaving of frequencies and polarisations if the beam coverages are to be adjacent (e.g. 3 beams on A pol F1, 3 beams on A pol F2, 3 beams on B pol F2, 3 beams on B pol F1) ? Will the customer earth terminals have to switch polarisations ? Maybe adjacent coverage areas will be covered by the next or previous satellites.

I've had another look at their webiste and I now realise that their market is cell phone companies and town/city ISP phone companies rather than home/small office/ cafe that I was thinking about. That way the earth stations can be far fewer, may be much larger, with both gateway hub and mesh connectivity, and link/power budgets that all make sense with comfortable margins.

Whetever, it all looks rather clever and I wish them well. It is good news for mid latitudes.

If the new system gets the price down to $500 per 1 Mbits that will be excellent. I know the typical price now is $4000 - $7000, but this is often based on very conservative assumptions, like small dishes and QPSK 3/4. If people really want low price per month now they can cut that cost by half if they would only use 16-QAM modems (e.g. CDM570L), larger antennas and linear BUCs. In certain rare special conditions (i.e. can see own signal and equal bit rates each way) then carrier on carrier is possible with yet further monthly price reduction - but with more hardware and skill cost.

Best regards, Eric.

I think the number "2133" came from that somebody said that the whole 16 satellites of O3b will be bandwidth equivalent to 2133 new transponders added to Africa. Probably meant standard transponders (which could really be different nevertheless..) I don't have the exact quote with me but pretty sure they won't have 2133 transponders on 16 satellites with each satellite weighing around 700kg.

Another important information is that with 16 satellites in orbit they will have 300 steerable Ka-band spots, each 500 km wide. They won't cover the whole world at once as they say. Probably these spots will be targeted to major cities in Africa or Middle East and will significantly drop prices on the market there. Other regions would still require traditional geostationary VSAT solutions as we have today or some sort of a terrestrial backbone. With $500/meg cost it definitely makes sense to build a chain of towers to distribute this bandwidth out of the spot rather than go for a $4000/meg geostationary solution.

And the last point, raised by a few people already, is what Ka-Band will turn into when it rains in Africa? Ka-Band shall be less tolerant to rain than Ku-Band. I know it can fall back to lower order mod when conditions deteriorate, however only real life will show how much of a meg there will be left once rain kicks in.

As far as I know they're planning to get 8 satellites up first and not all 16. Some time will have to go until we see O3b in full action. Also, somewhere there was info they're going to cover +/- 5 degrees from the equator first. I am not really sure if this is a true fact.