Hacker News Reader

Comment by Scubabear68 | Hacker News Reader

One thing I always wondered is how the hardware that compares the N computer’s output (and fails it silent if one has a different result) is itself hardened.

Failure of that I imagine would be catastrophic.

View on HN

Two links, to start your rabbit hole for today: [1] [2]

[1] https://en.wikipedia.org/wiki/Triple_modular_redundancy

[2] https://en.wikipedia.org/wiki/Byzantine_fault

I don‘t think that is what was asked for. The question is really what about the combining unit itself? And in my opinion the answer is perhaps that this unit is very simple and failure would be comparable to physical failure of the output wire or device being controlled.

I tried the search term "fault-tolerant voter circuit NASA". The voter circuit itself is a single point of failure and I guess that NASA designed that to be fault tolerant.

Figure 2 in this 1969 NASA report shows a TMR system with three voters but I'm not sure how that works

https://ntrs.nasa.gov/api/citations/19690031505/downloads/19...

One approach is to use powerful and cost efficient COTS hardware for generating the outputs, and then you can have a much simpler rad hard computer take those in to generate the final decision. You also need all kinds of watchdogs to make sure you can reboot if something shits the bed, and you have to be careful about how things communicate to make sure that one faulty computer won't spam the others into uselessness

Reply on news.ycombinator.com