perhaps this is more appropriately called a mixnet emulation rather than simulation because we are not using an approximation of our mixnet AQMs but rather the actually katzenpost server code.
the idea here is to:
gather a client send schedule from the real world
write a mixnet emulator that can use this schedule
iterate through various mixnet decoy traffic configuration permutations and analyze the traffic as a global adversary, perform various statistical disclosure attacks such as short-term correlation and long-term intersection attacks or measure the effectiveness of such attacks for different settings. use a shannon entropy model to describe adversaries uncertainty for each decoy traffic configuration permutation.
perhaps this is more appropriately called a mixnet emulation rather than simulation because we are not using an approximation of our mixnet AQMs but rather the actually katzenpost server code.
the idea here is to: