Re: [ppsp] questions about merkle hash tree mechanism for integrity protection of content between peers

[Arno Bakker <[e-mail]>]

On 05/11/2012 18:41, 邓灵莉/Lingli Deng wrote:
> Hi Arno,
>
> Thank you for the impressive speech and demos this morning.
>

Hi

thanks, but that was my boss, Johan. I was virtually present in the meeting.

> I have a few questions about the security proposals in current peer
> protocol draft:
>
> 1)Swarm ID, be it the roothash of a merkle-tree in video-on-demand cases
> or a public key of the initiator in live cases, is not straightforwardly
> inferred to a joining peer, which means there has to be some way to
> securely and uniquely derived from the common knowledge of a joining
> peer (for instance, a few vague key words to search what the user
> wants). I understand that, the pollution issue concerns not only serving
> as you asked, but also serving as you expected. What do you do in your
> implementation? Or the user just know what to check in the first place?
>

I'm not sure I get what you mean. The swarm ID is the identifier of the 
swarm, so if a peer wants to join a swarm it must know the swarm ID. As 
the PPSPP swarm IDs are self-certifying, a peer is always guaranteed to 
get what he asked for. The assumption is that the peer gets the swarm ID 
from a trusted source, e.g. a webportal that says "This BBC Top Gear 
episode that you want to watch has roothash X".


> 2)For ECS proposal, I doubt the one-to-one encryption negotiation is
> scalable for a peer serving more than one peers at a time. Since you are
> not targeting DRM requirements, revocable group keys may suite your
> needs, while the data can be encrypted once and for all given the group
> membership remains unchanged.
>

Dusan?

CU,
     Arno