According to the New Scientist (Hide and Seek, May 2001), a company called Invicta Networks had already perfected something similar to my adaptive IPv6 network over a decade ago. The principle is the same: the IP addresses on the network will constantly change, making it very difficult for an attacker to enumerate the network and identify potential targets. While I feel like a total gimp for not being aware of this before, it’s evidence the idea’s workable to some degree.
Invicta Networks claims this as a commercial service/product, but I haven’t seen examples of it deployed in the real world. I imagine it would cause all kinds of problems with stability, pen testing and diagnostics on an enterprise network.
But if the addresses at both points are always changing, nothing can communicate, right? It’s one of the reasons remote access tools require C&C servers with the current IPv4 scheme. Actually they can, if there there was a method for all clients in a group to know in advance what each others’ addresses will be, or of synchronising changes. Invicta Networks does this with something called ‘Variable Cyber Co-ordinates System‘, involving PRNGs and custom-built Network Interface Cards. Clients must send each other advance notice of address changes over encrypted connections, which again is something along the lines of what I thought of, but in my case the idea is to reduce the risk of discovery further and have the IPv6 addressing handled by an ‘Adaptive Network Algorithm’ and distributed initialisation value.