I had an interesting conversation last night with a nanosatellite engineer about how much computational power is required in a nanosatellite constellation and if it's necessary.

I had an idea or a vision for an experiment with decentralized technologies like blockchain and etc. and I was asking him for some advice or opinion. My opinion is that - implementation of decentralized technologies in the computer group to share the computational power has proved itself, and first decentralized global supercomputers are being developed (see Golem Network). They are primarily used for complicated computational tasks, such as 3D rendering, machine learning etc. Applying the similar concept to a group of satellites under one platform will vastly increase the possibilities of the data processing. Peer-to-peer sharing of computational power not only increases the possibilities of data processing of the orbital equipment, but also allows more complex utilization or processing of the payload, be it the complexity of raw data processing or the direct use of A.I. during the processing or control. Constellation like this is allowing us to create completely new kind of commercial hardware/software with limitation only by our creativity.

He had a completely opposite opinion. He claimed that computational power in nanosatellite does not matter and the whole trend in nanosat development is going another way.

I'm really curious about your opinion. Looking forward to your comments. Thanks

  • 1
    $\begingroup$ I can't think of any situation where a large number of cubesats spread out over tens of thousands of kilometers, each very expensive to launch and deploy, with perhaps only a dozen watts of solar power each, spending almost half of their time in darkness every 90 minutes would solve some problem that a room full of the same computers in someones basement couldn't solve faster and for a dramatically lower price. Can you add some discussion of what kind of problem or application this might address? $\endgroup$
    – uhoh
    Nov 17 '18 at 10:58
  • $\begingroup$ Let's say satellite imagery. Using satellite imagery as an oracle for instance, insurance applications. Distributed computational power in a decentralized network could be used to power A.I. to analyze what satellite "sees" on the ground, forwarding this data to insurance applications. $\endgroup$ Nov 17 '18 at 11:15
  • $\begingroup$ Or another example is autonomous vehicles as another sector with the vast potential for using satellite technologies. As with GPS, the satellite view on Earth is a fundamental part of an efficient autopilot. The functionality of an autopilot cannot be based solely on its A.I.’s internal sensors, but the view from the above is also very important for it to understand the objects surrounding it, and it adds to the safety of the passengers. By using the satellite technology, the autopilot can not only predict but also actively avoid existing or potential accidents, traffic changes and road blocks $\endgroup$ Nov 17 '18 at 11:15
  • 1
    $\begingroup$ Sorry :) yes, I'm thinking of decentralizing computers inside cubesats. $\endgroup$ Nov 17 '18 at 11:41
  • 1

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.