Starlink’s Inter-Satellite Laser Links Are Setting New Record With 42 Million GB Per Day

Slide from the SpaceX Starlink presentation on mesh routing via the laser links. (Credit: PCMag/Michael Kan)
Slide from the SpaceX Starlink presentation on mesh routing via the laser links. (Credit: PCMag/Michael Kan)

Although laser communication in space is far from novel, its wide-scale deployment as seen with SpaceX’s Starlink satellite internet constellation has brought the technology to the forefront like never before. This was quite apparent during the SPIE Photonics West event on January 30th when [Michael Kan] and other journalists attended a presentation by SpaceX’s [Travis Brashears] on the inter-satellite laser communication performance that was first enabled with the Starlink v1.5 satellites.

Among currently active inter-satellite communication systems, Starlink is by far the most numerous and with the highest bandwidth, reaching over 42 PB per day across its over 9000 space lasers (yes, that is over 9000) for a 5.6 Tbps throughput. Since these satellites form a mesh network with their 100 Gbps laser transceivers, a big part of using it efficiently is to route any data with the least amount of latency while taking into account link distance (maximum of 5,400 km), link duration (up to multiple weeks) and presence of other Starlink satellites before they become within reach. With this complex mesh in LEO, this also means that a very high uptime can be accomplished, with a claimed 99.99% due to rapid route changing.

For the future, SpaceX has plans to not only keep upgrading its own Starlink satellites with better laser transceivers, but to also make them available to third-party satellites, as well as beam the lasers directly down to Earth for ground-based transceivers. The latter is still cutting edge, despite it being tested to beam cat videos to Earth from Deep Space.

44 thoughts on “Starlink’s Inter-Satellite Laser Links Are Setting New Record With 42 Million GB Per Day

        1. There is when you have a limited power budget, solar panels are vastly inferior to what small nuclear reactors could do. Get safe and reliable nuclear reactors in space sorted out, with zero risk of a re-entry*, then come back with “there is no such thing as too much bandwidth”.
          *this bit is trivial for interplanetary missions ofcourse, and we could have nuclear powered interplanetary misssions already if governments and corporations would spend properly on space exploration instead of wasting “our taxes/profits from our purchases” on ever more intrusive bureaucratic meddling in to every area of life

        2. There are so many reasons that this is not the case. If there’s no need for the bandwidth, the added power consumption, system and software complexity, increase in system knowledge and pointing requirements increases the number and cost of sensors required etc etc. In space we don’t launch massive over-complication and over-provision, we launch what’s needed and what is within budget, then look to the next generation – it’s how you don’t get more spectacular failures and waste.

          As for Solar vs Nuclear – well that’s another argument, and ignoring the massive safety discussion, the power density and system complications of it (again) are not simple. There’s many many reasons they’re not more common, not just the safety aspect.

          1. From a space engineering standpoint, I totally agree with you, but from a space science standpoint – you really do always need the bandwidth. A lot of space mission planning is the science side saying “plz can we haz more” and the engineering side saying “no, you got your cookie, eat it slowly.”

    1. He just came up with that nonsense on the spot, in reality to walk you need sensory feedback, and you won’t have that with neuralink, so you need an exoskeleton and gyros and what not then pretend that a neuralink does anythinng more than some electrodes to the body like they do now with the same result.

      And even all that will take decades probably.
      So no you won’t have paraplegics walk and the blind see any time soon with neuralink.
      Don’t believe silly impromptu salespitches.

    1. If this statistic was true at one time, it’s radically changed (along with everything else) in the Netflix/youtube era.

      A normal day-average number for consumer facing ISPs is ~60% ‘video’ (that is, non-porn video) these days and adult is somewhere down around 1% – both percentages being taken off the total coming in through the edge of the network.

      Source is unfortunately close to ‘dude trust me’ but I work on systems for internet traffic classification. (Amusingly ‘adult’ traffic is pretty easy to classify because it’s more frequently in specialized hosting rather in the whole AWS blob)

    2. There are more important things than p*** in internet traffic, many of them needing just as much privacy and anonymity as that does. I think it is about time that Musk showed he really cares about the internet, and started using Starlink to provide uncensorable, unsurveillable internet connectivity to oppressive countries round the world. Those countries, even those with anti-satellite missiles, would be powerless to intervene, because none could dare try to bring down Starlink satellites lest in doing so those countries unleashes a Kessler Syndrome storm which would tear up all their expensive LEO military comms and recon satellites too. Musk, you know you can, you know you should, you know you want to, you know that censorious states are powerless to down your sats… Starlink as an anti-censorship tool NOW!

      1. The internet itself is unreliable and insecure. Reliability and security are provided by the users (clients) who retry and encrypt. The to and from addresses in internet packets and normally public and countries ‘censor’ by blocking destination addresses.

        Starlink merely needs to provide an internet that has no blocking and it’s a major win for freedom.

      2. Starlink, all telecoms, must meet certain requirements to operate in each country. These are all over the board. Starlink cannot operate in South Africa because the company is not 30% owned by historically marginalized groups. Many countries on the more authoritarian end of the spectrum require the ability snoop on internet traffic. Just because it is satellite-based does not mean Starlink does not need to follow each countries rules.

    1. But it’s not him, it is the people that work for him that are doing all the clever stuff, same goes for X (what’s left of it), Tesla and Neuralink.
      At least they seem to be keeping him away from the day-to-day running of those companies. X is an unmitigated plane crash but SpaceX seem to be actually making some pretty decent stuff. The Neuralink stuff is a bit scary and Tesla have got problems but still seem to be selling cars so…

      > Isn’t it amazing what a smart, s/charismatic/egotistical/ megalomaniac can do with tens of billions of dollars of other people’s money? God bless unfettered capitalism!
      FTFY

    1. That’s not Musk’s problem. That’s for the rubes that buy the company when it goes public. The original SpaceX investors make out like bandits from the sale, and SpaceX continues to get the replacement launch revenue, but the new mom&pop investors are left holding the bag for maintenance.

    2. Citation needed. Historically its been 1,000 a year to get the system up fully (5,000 since 2019, 12,000 projected total). They are not losing 10,000 a year if they are supposed to stay up for 5-7 years, realistically you wilo lose them periodically, near the rate you put them up. Many may also keep working for longer after the early failures drop out.

      1. My mistake, I got a 50 000 satellite full constellation number from somewhere when the actual one seems to be that 12 000. At 5 years satellite lifespan that comes down to 2400 sats a year for maintenance, which is still quite a lot.

        1. And 2400 sats per year is one Starship launch per month. A nice continuing revenue stream for SpaceX, and a continuous drain on the future retail investors of Starlink. To the tune of a gigabuck per year… This all only works if a substantial fraction of Earth’s population subscribes to the scheme.

    3. The entire original argument for Starlink or any satellite constellation is that LEO satellites with laser links can conceivably get latency below hard-wired connections b/c of the slower speed of light in the fiber. I don’t know if they’re on track for that, but if they are, it’ll definitely be profitable.

  1. My. Dream is for. Starlinkt to. quickly break the big 4 cel phone monopoly we are currently saddled with here in the USA.the fees are outrageous compared to ours here in Indonesia..

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.