The telemetry and tracking itself for the laser links. All the satellites need fairly accurate orbital information from all other satellites; there will be several thousand satellites capable of performing autonomous orbital adjustments to avoid debris and for their own station keeping. There has to be a convenient and reliable fallback for automatically syncing satellites with the rest of the network after a reboot or loss of communication.
The packet network needs to maintain an efficient routing table and provide low packet loss to end-users. My guess is that the only practical choice will be establishing a large number of point-to-point channels with their own retry/error-correction management. TCP/IP isn't capable of dealing with more than about 1% packet loss and I doubt raw optical/wireless links will be reliable enough especially with topology changes, requiring extensive store-and-forward hardware to buffer transmissions waiting for a reliable route or de-duplicating multicast packets at the receiving end. Reassembly and de-duplication on receipt is theoretically feasible; if a tight reception window can be maintained then specialized hardware could filter duplicates sent over 2 or 3 redundant routes to a local receiver without increasing observed latency or using unreasonable amounts of RAM. A combination approach could allow a tradeoff between using double or triple the raw bandwidth but providing low-latency reliability vs. higher latency with retries while maximizing throughput depending on the current network load. It could also be hardcoded for different traffic classes.
I'm not involved in this problem space in any technical capacity but it sounds like a very fun set of problems to solve.
Accurate orbital information might not be necessary or possible, if you can perform broad scanning that can quickly lock in your given target, but it certainly doesn't hurt. The issue is that even having an orbital track, because of the sats being so low in LEO, atmospheric drag will change your orbit rapidly enough that this information can get out of date quite fast.
Convenient and reliable fallbacks I feel are largely a solved problem for SpaceX. They've built their Starlink bus by reusing a lot of the software from the Falcon 9, Dragon, and Starship programs that already had to handle even greater levels of reliability.
Routing and retransmits are indeed quite a novel area for this kind of service. But the laser link will only be locked on one other peer satellite at any given time, and I don't think they plan on reorienting the sats in order to establish laser comms because of the effect it would have on drag as well as albedo. The former impacts service life per sat, and the latter has been a big rallying cry for Starlink opposition due to the impact it has on astronomy and the visible sky.
So if you likely can't reorient to retransmit, your only options are the peer laser link or the ground transit exit node, either or both of which might not exist, but I feel like you would just ack the packets on each hop to your peer and leave it at that.
http://www.circleid.com/posts/20190906_inter_satellite_laser...
While a major leap, Starlink could potentially service internet backbone traffic at substantially lower latencies in such a world.
You can control the overhead while increasing recovery rate over multiple paths fairly well, with non-integer overhead (>1x but <2x, etc) with erasure coding techniques. Some erasure codes can allow you to control that overhead (code rate) precisely, but have legal imbroglios: https://en.m.wikipedia.org/wiki/Fountain_code
Having said that, I don't think this situation will apply to spacex. They probably won't get random bursts of errors. Noise levels will be very deterministic and can be calculated ahead of time by looking at what stars are "behind" the satellite.
I'd love to be on the Starlink team, they are building some really cutting edge stuff. There are not many places or times where you can have such a big impact on the world and their team happens to be one of them. Good times.
[1] https://www.sciencedaily.com/releases/2020/05/200522095504.h...
They're incredible :)
Then there's the issue of signal attenuation at distances like these, along with the power budget from the solar panels to ensure that you aren't burning all your power just on laser comms.
I can't imagine these links would be anything more than 1:1 at any given time, at least not at first. Maybe later they might be able to handle simultaneous laser links from 1:3 or something like it, but I highly doubt that's their current capability.
Even some of the best latest ground-to-sat laser comms links are on the order of ~7 gigabits per second. [1] I imagine those are far larger sats than Starlink, and their entire power and thermal budget is likely spent on comms. And the ground stations can be orders of magnitude larger and power hungry in relation to a Starlink sat. Note also that this is from GEO, so you aren't likely having to handle substantial relative movement either.
[1] https://en.wikipedia.org/wiki/Laser_communication_in_space#2...
Satellite-to-ground optical laser links are another matter, but you can always route a few hops around weather systems.
https://www.spacex.com/careers, category "software development".
But I do see on the careers page there's now a (single) Palo Alto based Software Engineer position open.
Intriguing!
The second challenge is that I am in a position to be reasonably frank when the biggest threat one can hold over me is to be fired. I have found that some managers don't appreciate employees for whom their "hire/fire power" over them doesn't give them any leverage. This combined with a general perception that I am a good manager, means that I tend to develop a followship whether or not I'm in a leadership role. This can lead to "issues" no matter how hard I try to prevent them.
Or does the "eligible to obtain the required authorizations from the U.S. Department of State" apply?
If I were you, I'd read up on ITAR qualification and, if you think you would be eligible and are highly qualified, just apply.
and also relatively overhead of the CPE antenna, since the starlink customer terminal is a phased array that does dual beamforming in a 'cone' of view directly above it. just because a satellite is visible 5, 10 or 15 degrees above the horizon from the POV of the CPE doesn't mean it can talk to it. the system is definitely reliant upon a fairly high satellite density.
with the ability of the satellite that's generally overhead of a starlink earth station to talk to the satellites immediately behind, and preceding it in its orbital plane, and then those two additional satellites to talk to the CPEs underneath them, the possible coverage area can be greatly increased.
3D visualization of starlink orbits and coverage footprints:
Right, compared to reasonable in-ground ISPs.
I think Starlink sat to Starlink sat links will also help bandwidth, specifically in the case where a given downlink connection is heavily utilized, the system can shunt traffic to another slightly less ideal but less utilized ground station.
> unlocks a very financially lucrative consumer use case
That's a bingo! I think there's a pretty good chance that this could turn into a pretty epic cash cow for Starlink.
it's very low data rate but also guaranteed lower latency than the submarine cables.
Maybe with that out of the picture other uses like telepresence for remote surgery can edge in. I've been waiting for a while for that to really take off seems like an awesome way to provide services you couldn't afford to to remote places.
I have read that Starlink, given sat to sat relays, should be able to beat any possible ground based system as long as the distances are great enough. Does that match your understanding of this?
Starlink is at a disadvantage because fundamentally it's pathing across a sphere with a larger radius so even when the path is direct across the Starlink cluster it still has to traverse a longer distance than the equivalent path on the ground. Adding to that the path won't always be directly across the surface of the constellation, depending on where you are and where the destination is you'll find some pockets where the link has to zig zag across the constellation slightly adding to the additional link distance. This [0] isn't necessarily a perfect representation but it does show how starlink's linking may work and how it's not a perfect net. It also doesn't account for the acquisition times for new laser links which were on the order of 20 seconds in another laser sat link test.
[0] https://youtu.be/AdKNCBrkZQ4?t=103 (it's also comparing fiber to Starlink where the best in class now is radio links generally)
> where the best in class now is radio links generally
What would I put into a search engine to learn more about these more recent long distance radio links? Thanks!
There are atmospheric and practical effects that make going more than 50-60km with PTP microwave, and data rates above 700-800 Mbps, increasingly costly as the distances increase.
It is totally possible technologically to build very long chains of many microwave PTP radio hops, as some high frequency trading companies set up between new york and chicago, but the aggregate throughput and capacity is minuscule compared to fiber.
Would be happy to answer any more specific questions.
Here's an example of a current state of the art point to point microwave radio, for use in the common FCC licensed part 101 bands (6, 11, 18, 23 GHz, etc)
https://aviatnetworks.com/products/all-outdoor-radio/wtm-400...
if you google 'starlink train' and look at some videos, when a batch of 60 starlink satellites is launched ,they're all at the same orbital inclination. they remain at the same inclination but as the weeks go past after their launch, they are spread out to follow each other at several hundred km spacings. but they're still following each other in what is basically a strung out conga line of satellites. communications between forward/rear satellites in the same batch should not be nearly as difficult to aim.
what would be difficult to aim and maintain links on would be cross-plane links between two different sets of satellites, with very high differing relative velocities.
I dug up the paper with the technical details.
I did not find anything suggesting it'd be used productively for the satellite if it manages to deliver. Do you have anything about that you can tell/link?
I'd be curious what the initial design weight was, before the thermal expansion problem that caused the aluminium block.
As for the data rate, it seems a mechanically simplistic fold-out mirror (released during deploy, say with a current pulse into a shape-memory-alloy torsion spring to delay the unfolding until after the ejection, or just a simple friction break and a normal torsion spring) could significantly decrease the beam width when made from e.g. zerodur and fabricated as an offset parabolic dish. Slightly modifying the refractive collimating optics of the current design should make that approach possible.
I'd expect vibration in space to be a negligible issue, so an edge-mounted mirror should be stiff enough.
https://elib.dlr.de/135960/ https://elib.dlr.de/135960/1/1150604.pdf
In GEO that's not a problem - a lot of the planet is in sight. You have a ground station in New York and you can bounce off a satellite over the Atlantic and land the signal in Nigeria just fine.
With Starlink the orbits are really low, so the distance to the ground station is low. That's fine if you are in the backwoods in Washington and bounce to a receiving station 100 miles away in Seattle, it's no good if you're at sea, or (in this case) at an Antarctic station -- one which can't even see GEO satelites.
it's the pole that has problems seeing geostationary, and has up until very recently been totally dependent on 7-8 hours of coverage a day using big tracking antennas to talk to old geostationary satellites that have wandered somewhat out of their original orbits, so that from the POV of the earth station at pole, they can be communicated with part of the time.
Most data and internet from pole is via ~4 hours of DSCS a day, ~4 hours of skynet (slowww...) or when TDRSS decides to gives South Pole time (but in small chunks of time).
I usually try to ssh in my experiment at Pole the start of DSCS time, which unfortunately can be in the middle of the night for me.
But there has been nowhere near enough money available to do something like set up a pair of molniya orbit satellites with long apogee dwell times over the center of antarctica.
Assuming that the topology will be additional highly-inclined/polar orbit starlink satellites, communicating with each other in a laser chain, this will be dependent on starlink setting up earth stations near fiber somewhere in southern Chile, Argentina or in South Africa. Or south australia, tasmania, new zealand.
Now that I think about it though, the Starlink user terminals likely won't be able to survive the harsh winters at the poles. -30 deg C and low tolerance for high winds means this would be tough unless some type of non-obstructive sheltering could be provided.
Even if you maintained multiple $1.25 per minute (or $5 per megabyte) Iridium links, operating 24x7x365, it would be a teensy tiny drop in the bucket compared to the total budget of the station.
https://www.wolframalpha.com/input/?i=2*6300*arccos%286300%2...
I have seen no evidence that traders will be the primary users. Do you have any evidence or cases where they are doing or planning this. Are you just guessing?
Starlink gives them the same opportunity across Atlantic, Pacific, and Indian oceans. You can bet that there will be orbits that exactly link London and New York, New York and Tokyo, New York and Singapore, London and Singapore, etc. Traffic not sent under extra-high tariffs will be artificially delayed enough milliseconds to match fiber.
It would not be surprising if these low-latency contracts, for traffic amounting to well under 0.1% of total capacity, provide 10% of revenue.
It would also not be surprising if future satellites have tens of TB of storage onboard to proxy streaming for the highest-demand shows of the moment, and broadcast capability to multiple ground stations simultaneously for live streams, particularly soccer matches.
Most likely, aside from financial transaction data, routing will always offload packets to ground terminals as early as possible. I strongly doubt zigzag routing will happen at all.
Another may be gaming.
Another is as you say planes and ships out of reach of ground stations (very sparse situation however).
The financial upside is clear, but what kind of emergency capabilities do you mean?
But I too was interested to see if they had an actual idea in mind?
I would think the computer game market. Maybe other markets that have high end mature users where client side is very good so lower latency might be something that matters relative.
At some point, the communication has to enter the terrestrial network. That entry node is likely to be under the influence, if not outright control, of Five Eyes security infrastructure.
You would need to build out a lot of infrastructure to escape the sort of ubiquitous surveillance that is available to those organizations.
Not necessarily. If you're communicating between two SpaceX terminals, that could feasibly avoid SpaceX ground stations. I don't know how capable SpaceX's routing actually is to know if this is within their scope though.
If it's flying through the air and somebody really wants to grab it they will, whether they can read it depends on how good your crypto is.
LEO is far from a true vacuum. But it's sparse enough that you have limited beam scattering. Getting a read on a sat-to-sat laser emission from the ground is nontrivial.
That said, I imagine that for any serious bandwidth, you won't get enough SNR to decode anything from the ground.
I guess if you got an observation platform up closer that would help.
If they were strongly encrypted there may be no reason for a nation state to observe the intersatellite links, but I wouldn't expect this at least for the near future since Elon's focus will probably be to cut corners to get the absolute best performance possible.
[0] https://en.wikipedia.org/wiki/Consultative_Committee_for_Spa...
[1] https://www.google.com/search?q=ccsds+protocols&source=lnms&...
Edited to add: I'm not making any claims on what protocols Starlink uses specifically. Just describing the general approach.
In practical terms I don't think it's a major concern. Much more likely that the spy agencies would tap the lines coming out of the ground stations.
Which is a significant step backwards for e.g. Russia.
Currently, all data between Asia and North America runs on cables. Anybody can spy on those. If those data did satellite laser hops, only those with ground stations would have access. Everyone else gets locked out.
I have no doubt that (like every other geo-mobile satellite constellation) the system allows for "port mirroring" on point-to-point links, or (more likely) relays everything through a ground station even when doing so has performance implications. This allows them to fulfill their obligation to government(s) surveillance.
They can't (as far as I know) use beamforming with lasers, so they can't be steered to different terminals as quickly.
As the sister comment implies though, Starlink sats are in low Low-Earth-Orbit (LEO) so they experience non-zero drag from the atmosphere. This slows them down, causing them to gradually fall back to Earth, creating a natural expected service lifetime for each satellite in orbit.
So the expectation is that SpaceX will have to continually put up more and more replacements indefinitely as older sats decay and burn up in the atmosphere. These 10 laser sats are just the latest among a string of sats that are yet to come online with newer and better capabilities.
(downvotes don’t change facts, kiddos)
The value that sat-to-sat laser links provide is that they create a low latency, high bandwidth path that stays within the Starlink satellite network. Before these 10 satellites, each Starlink satellite has only been capable of communicating directly to ground terminals (either consumer, transit, or SpaceX control). For traffic that is intended to move large geographic distances (think transcontinental), this can require several hops back and forth between ground and space, or the traffic from the user terminal is exited at a node that is geared for transiting traffic and most of the data transits along existing ground Internet links.
By performing this type of transit directly in space, and exiting at a transit node nearest the destination for the data, you greatly reduce latency. Bandwidth still might not be great, but what this does is unlocks a very financially lucrative consumer use case: low latency finance traffic and critical communications. There are many use cases around the world where shaving even 10-20 milliseconds of latency on a data path can unlock finance and emergency capabilities, and this is a long fought battle throughout the history of these industries. As an example, if you got a piece of news about a company in Australia, and wanted to trade on it as quickly as possible in USA, if you can beat your competitors by 10-20 milliseconds, that can mean a lot of money.
Laser comms for Starlink sats have long been planned, but have historically proven to be quite hard to get working. They also depend on a sufficient critical mass of satellites so that a given sat actually does have another sat within lock to send the traffic towards.