895583 wrote:Depends. A real-time intercept might be challenging, but someone who is intent on determining strategic intent from the conversations might collect the packets and write software to attempt re-assembling them based on the digital-analog conversion of the voices. While bits in the digital stream of disjointed packets cannot be distinguished from each other, the tone, pitch and audible level of sound can, when it is converted to a wave form. Software could match these properties at the ends/beginnings of each packet, and based on matches, attempt to string them together.
The net effect of what I describe is intended to be a jumbled (if you like) sequence order that is itself encrypted (via SSL). Assuming that SSL does its job, and protects that ordering, it would appear to be onerous and irksome for an intercept to re-render the packets (VoIP) into a meaningful or intelligible stream.
In practice, this scheme likely wouldn't work particularly well for voice packets, given that they are highly perishable.Once again, it depends on the context and/or the people on each side of the conversation. Even if it takes an attacker 10-days of processing to string conversations together, it might give them valuable information on anything that has a shelf-life of longer than 10-days.
895583 wrote:Have you tried the same with non-linear function(s) with logistic map to improve such?
Hello, and hope all you fine first class passengers are in good fettle today. Especially EJP if he's around.
Now I have an operationally acceptable Java VoIP system, I am considering the sagacity of using a randomiser to re-order the packets (since they are UDP, SSL won't help me here) to pseudo-encrypt the stream. I wondered if anyone would like to offer a comment - of any kind - about the pros and cons (if there are any) of implementing such a thing. Thank you gentlemen.