[rc5] Suggestion for protocol

Tim Charron tcharron at interlog.com
Tue Jun 24 22:30:22 EDT 1997


> >Somebody here mention Zero knowlegde proof and my suggestion was based
> >on similar ideas. You do not let the client know what part of his work
> >will be checked, so it needs to do all the work.
> 
> How about this, with every key tested there is some result for the 
> decryption, a running total or checksum is calculated on the entier
> keyspace assigned. This checksum is reported back to the server. The
> server passes out known keyspases at random intervals and verifies
> the returend checksum. This could also be used to double check any
> keyspace by sending it to a second computer and see if the checksum
> matches. 
> 
> This way if for any reason a client is suspected, checksums of the 
> returend keyspace will authenticate the result.
> 
> The server does not need to calculate the checksum, it can take results
> from a known and trusted client and pass them out to the world.  And by
> keeping the checksums and where they came from, the spammer can be 
> tracked down and be tared and fetherd :-)

The problem here is that the server needs some way to calculate the 
checksum.  If you're calculating a checksum twice, then that's a lot 
of wasted cycles, no matter how you go about doing it.

There's a simpler way.  When a client receives a block, it's checking 
2^28 different keys.  Instead of doing a checksum on all of that, 
have it do a checksum every 2^18 keys (for example).  There would 
then be 1024 different checksums.  The client is required to 
return the low 2 bits of each checksum (2048 bits, or 256 bytes).  
The server then randomly picks one of these 1024 checksums.  It then 
calculates the checksum for that block (or delegates the task to 
another machine).  Any client not doing all the required work is 
pretty likely to get caught after a handful of blocks.

Tim Charron
tcharron at interlog.com
tcharron at ctfinance.com
http://www.interlog.com/~tcharron/

----
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.



More information about the rc5 mailing list