Re: Generating a large sequence of unique, random numbers

From: Kevin Buhr (buhr_at_telus.net)
Date: 05/30/03 

Date: Thu, 29 May 2003 23:08:56 GMT

Kevin Buhr <buhr@telus.net> writes:
>
> The main attacks against this serial number / verification hash scheme:

Oh, I guess that in a world of poorly secured web servers, it's stupid
not to point out:

4. An attacker who can compromise the web server to determine the
    secret key K can generate valid codes with it.

It would be nice if there was a scheme that included a public key
cryptography component---the private key would be used to construct
the 53-bit numbers, the "public" key to verify them. The public key
would be protected on the web server but, even if it was compromised,
an attacker could not use it to generate valid numbers only to check
the validity of existing numbers.

The problem is that I don't know of a public key cryptosystem that
can be made secure using a block size of only 53 bits.

A technical solution to make attack (4) more difficult would be to
have the web server(s) verify keys by passing them to a more carefully
secured system (one that does nothing but accept a code, verify it,
and return the result of the verification). 

-- 
Kevin <buhr@telus.net>

Relevant Pages
Quantcast