[tor-dev] Effect of padding on end to end correlation false positive rate

[coderman]

On 10/20/15, grarpamp <grarpamp at gmail.com> wrote:...
>> [1]: https://blog.torproject.org/blog/traffic-correlation-using-netflows
>
> My thought was solely restricted to analysis of network
> traffic by *passive* adversary... not involving any collusion
> by actives over circuits they can see inside or pump within
> any given onion layer, though clocked and checked network
> fill by all proper nodes would inhibit pumping by actives.
> I talked on list with someone at briarproject and wherever
> else on idea of filling the network with traffic vs the passives.
> Apologize for not making time to review Mike's proposal
> or develop further talk yet. Someone will review / integrate
> fill padding of network with regard anonbib, Mike's, etc I'm sure,
> as it is clearly (to me at least) a weakness of non-filled
> non-store-and-forward networks vs the passives which we
> all know and love.


note that userspace stacks with multi-homed mobility-capable IPv6
ORCHIDv2 addresses can maintain logical connectivity over a set of
paths/circuits such that active attacks reduce capacity but are unable
to execute the easy confirmation and tagging attacks available in a
stream oriented implementation with TCP links over IPv4. in theory...
;P

best regards,