You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Here it is informally explained that in Whisper the origin of a message could be found out by a powerful adversary.
Perhaps the next issue is obvious and known and not the goal Whisper, but I assumed differently so I'll post it here anyhow. Also, I've not actually tested it.
A Whisper envelope contains the expiry and ttl fields. Typically in implementations: expiry - ttl = creation timestamp.
Now I question if a peer can not simply abuse this timestamp to figure out the origin of the envelope.
A peer his direct connections, could have a fairly stable round-trip time (RTT). Just looking at the timestamp could perhaps identify the peer as a sender or not. Especially over multiple messages.
This is however somewhat more complicated:
First there is the PoW calculation which will delay the queuing of the message. This delay will also have some randomness. However, the delay will probably not be enough compared to a usual RTT/2. Especially if the selected PoW is low (e.g. very low setting of PoW in Status).
Secondly, messages are send in batch and not immediate. This is not a requirement of EIP-627 but it is generally assumed and done in geth and nim-eth. Depending on the send interval (300ms currently), this will make it more difficult to figure out the origin of the message (1 - 300 ms can be added as delay).
In the worst case (e.g. messages with less delay added than RTT/2), it is likely that some messages will be able to be linked to a peer you are connected with.
Combine this e.g. with a node monitoring the messages of some most used public channels on Status and you could perhaps link IP-address to the messages & Status names of a user that this node is connected with.
Of course these are public channels but still I believe pseudoanonimity should be a goal even on those channels.
Possible ideas of improving this:
add more time randomness by decreasing expiry with random amount.
add more time randomness to the batch envelope send.
TBI
The text was updated successfully, but these errors were encountered:
Here it is informally explained that in Whisper the origin of a message could be found out by a powerful adversary.
Perhaps the next issue is obvious and known and not the goal Whisper, but I assumed differently so I'll post it here anyhow. Also, I've not actually tested it.
A Whisper envelope contains the
expiry
andttl
fields. Typically in implementations:expiry - ttl = creation timestamp
.Now I question if a peer can not simply abuse this timestamp to figure out the origin of the envelope.
A peer his direct connections, could have a fairly stable round-trip time (RTT). Just looking at the timestamp could perhaps identify the peer as a sender or not. Especially over multiple messages.
This is however somewhat more complicated:
In the worst case (e.g. messages with less delay added than RTT/2), it is likely that some messages will be able to be linked to a peer you are connected with.
Combine this e.g. with a node monitoring the messages of some most used public channels on Status and you could perhaps link IP-address to the messages & Status names of a user that this node is connected with.
Of course these are public channels but still I believe pseudoanonimity should be a goal even on those channels.
Possible ideas of improving this:
TBI
The text was updated successfully, but these errors were encountered: