Today it is very easy for an attacker to harm a remote network server essentially for free. The amount of resources needed locally by an attacker has no relation with the mobilized attack forces. We would like to invent an Internet physics that couples tighter cause and effect. That is, if a remote server should experience some force or pressure, this pressure has to be generated locally in a first place. Only by investing local energy or by doing local work should an entity be able to afflict the remote place with some action. That is, the Internet has to conserve virtual energy and impose some mechanics law across distance.
In this research project we propose to design and study run-time environments for networking protocols which inherently enforce desirable global dynamics. As a first target objective we look at TCP-unfriendly traffic. One approach is bottom up and is based on an artificial physics engine. The second approach is more top-down and envisages generic controllers which can switch among implementation alternatives to find optimal operation configurations. These two approaches are complementary and should be coupled. Common to both approaches is that we cast overall system goals as equilibria: Ideally, the forces which steer a system to its optimal point of operation shall be an intrinsic part of the run-time system and should not have to be (re-) implemented in each protocol again.
In this extension phase of the project we continue to use artificial evolution to let a communication system find optimal configurations (where undesirable behavior can be negatively rewarded) - a prototype is currently under development. In parallel and using analytical and simulation methodologies, novel media access protocols (e.g. access in wireless networks) are explored that show "physics-like" properties.