Hacker News Reader

Comment by mattrp | Hacker News Reader

just remember the underlying unit economics of LEO vs fiber. A repeated system is generally thought to be capable of 40+Tbps / pair where as LEO might be good for gigabit at best.

View on HN

I have no idea where you got that figure. GSO birds do can ~1Tbps already (EDIT: Viasat’s birds, for instance: https://corpblog.viasat.com/viasats-global-satellite-constel...). LEO is a factor of 50-100 closer. Inverse square law and all that means for the same aperture and power, you can get far better bandwidth.

GSO is very high orbit (22K miles) and therefor has much higher latency. The lower orbit (100-1K miles) constellations are lower bandwidth because of the high angular speed of travel removes the ability to do directional (high powered) transmissions.

Incorrect. The LEO constellations use phased arrays (or similar) and are able to slew at whatever angular rate you could want as they are electronically steered. Both for the terminals and for the satellite side. Iridium has used phases arrays in its LEO constellation for decades now, and the newer constellations like OneWeb and Starlink use phases arrays on the terminal side as well.

Are you talking based only on what is launched right now? Because most of the Leo satellite companies are working on point to point laser backbones which is definitely capable of 10gbps right now. And I don't know of any limitations of open space vs glass that would keep them from improving that to at least as much as we are currently able to get with glass.

It seems that most of the limitations of point to point laser right now are due to atmospheric conditions which is not a problem in leo.

40tbs is 100x400G.... Vs 10G. I agree that Leo will open up opportunities but it won’t supplant fiber economics.

it's the same underlying WDM tech - no reason why LEO laser comms constellations can't hit the same (infact can do more since FSO has less restrictions on wavelength bundling than fibre)

This makes the LEO networks sound more like the HFT microwave networks that are expensive and optimized for latency not throughput.

I am not sure exactly the distance to the new Leo constellations. Maybe 2k km? If that’s the case even with the speed of light advantage it’s still roughly 30ms round trip just to go up and down not including any electronic delays in the equipment. So unfortunately NYSE-LSE would still be faster on fiber.

Starlink is planning to deploy at 550km according to Wikipedia.

I recall some hacks to force a call via cable to get improved quality decades ago.

Reply on news.ycombinator.com