Thanks NASA, for trying so hard to keep my interest in the program. Maybe next time I'll not vote pro-NASA.
NASA boffins have diagnosed and corrected the glitch that forced nuclear-powered, laser-packing space tank Curiosity to rely on its spare computer. The rover is now using the spare, but the “A” computer is once again ready for duty if required. The craft has therefore resumed its analysis on the powdered samples of Martian …
I forgot this is the age of 5 second attention spans, where things only get funded if the result is a big explosion or the ability to slap an American flag down somewhere (in the name of Freedom(tm) of course).
Maybe the next rover should incorporate an automated tweeting service to splurt inane crap every five seconds?
"OMG wheelz r so drty LOL #newshoes"
And thanks to the average twatter intelligence level, garbled, corrupted messages won't even get noticed.
Or better, a proper MAC (which would also prevent people other than NASA from sending it any commands at all...)?
Seems a strange thing to have to worry about in such an otherwise advanced bit of kit. Should be simple to guarantee that messages either get through intact or not at all.
Even the best error correcting codes fail in practice when the signal is drowned out by MASSIVE random noise output by the sun. It would be like trying to shout at somebody on the other side of a football pitch while Motörhead are doing a gig in between. I think you will find that quite tricky (this is why people text each other in the disco, instead of talking).
Now make the football pitch a few hundred million km across, and replace Motörhead's few ten's of kilowatts of power by 3.9x10^26 Watts, and you get the picture.
The price of bandwidth doesn't enter into it. The equipment on Mars and on Earth, the distance between the planets and the position of the Sun fix the maximum total bandwidth.
If you really want round-the-Sun radio coverage, you shouldn't be looking at error codes, but at some relay satellites in Earth's orbit around the Sun, but lagging or leading Earth by a significant amount. Then between Earth and the satellites, one should always have line of sight to Mars.
NASA already use R-S and convolution-based ECC on their transmissions anyway; I'm just talking about some kind of hash on the end of each packet. ECCs can occasionally decode with undetected errors, but the chance of any kind of noise corrupting a packet and then also correctly adjusting the hash is FAPP zero.
Okay so we have Reed-Solomon FEC and hopefully some form of trellis convolution coding as well on the stream at bit level but where are the longitudinal checksums (CRC32, CRC48, CRC64) across the packets at block level and what about the message integrity checks (MD5, SHA1, SHA256) at whole file level?
Why has the B computer got it's own set of hardware if they never intended to use it? I suppose it's multiple redundancy; but it you're chucking this stuff millions of miles surely you want to use every available instruments as much as possible, and be able to control them from both computers, not lock it off in the "other half" of the system?
Actually, Shannon's theorem says you can do better than some arbitrary error rate on an infinite message as long as you include above a certain minimum rate of bit correction according noise (assumed to be random bit flips) on the channel. In effect, whatever error-correction coding scheme you actually implement, there exist more efficient ones. So, no, CRCs are not the ultimate coding scheme.
Biting the hand that feeds IT © 1998–2021