On 11/12/2015 10:25 PM, Steve wrote:
> Hermes was one of the first point-to-point over cellular modem ideas I 
> believe. The algorithm doesn't look that complicated, but I would 
> guess the difficulty is in the details. They say they could get 1200 
> bps through the phone, which is less than the 1600 bps used by the sm1000.
>
> As far as I know, jackpair has never been shown to work, and being a 
> year late there have been no examples of it working. They seem to have 
> blown all their money on a failed plastic injection molding for the case.
>
> http://www.news.cs.nyu.edu/~aditya/Research_files/dhananjay-mobicom10-hermes.pdf
>  
> <http://www.news.cs.nyu.edu/%7Eaditya/Research_files/dhananjay-mobicom10-hermes.pdf>
>
I took the time to read that paper properly. Very interesting. I'm quite 
surprised how well they were able to do. They didn't try using some of 
the toughest paths you will meet in the real world, but they did 
evaluate performance over a linear->AMR->[probably ulaw]->EFR->linear 
path, which is fairly tortuous. There's no frame syncing along a chain 
like that, so the signal can really be munged in ways that are not very 
predictable.

jackpair just sounds sad. After taking an age to get pretty polished 
hardware it seems they are only now doing extensive testing and finding 
things are so nice in the real world. This sounds like one of the more 
backwards development processes I have heard of.

Regards,
Steve

------------------------------------------------------------------------------
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Reply via email to