[Cryptography] traffic analysis -> let's write an RFC?

[Jerry Leichter]

On Jan 29, 2015, at 2:35 PM, John Denker <jsd at av8n.com> wrote:
> Here is a scenario.  This does not solve all the world's
> problems, but it has some value.  If nothing else, it
> proves we should not give up.
> 
> Suppose there exist cheap links and expensive links,
> e.g. local versus transatlantic....
Very nice analysis.

I wonder if one can re-purpose network coding techniques for this purpose? (Network coding is a technique designed - initially - to allow multi-casts to more efficiently use a network than appears naively possible.  The trick is that nodes in the network have more choices about what to do when they have multiple input packets that need to go over a single output link:  Rather than simply choose one packet at a time to send, they can send any linear (over Z/2) combination of the input packets.  http://iest2.ie.cuhk.edu.hk/~whyeung/tempo/Ho_Lun.pdf has an introduction - Figure 1.1 gives you the basic idea.)

The interesting thing about network coding which makes it potentially applicable here is that it can be built by forwarding *random* linear combinations of the inputs; each node sends along a description of which inputs you chose to combine.  If instead the nodes shared a key and, in synchrony, generated those random combinations deterministically, they wouldn't have to forward them - but an attacker without the key wouldn't be able to decode what was being sent where.  (Maybe.  Much analysis needed - this is just a wild-assed suggestion.  Assuming this hasn't already been done - I don't follow the literature on this stuff - and it can be made to work, it's at least a CS Masters thesis, perhaps a PhD.  Anyone looking for a topic?)
                                                        -- Jerry