Game theoretic model for IP routing
Kadri Tõldsepp

The existing network of Internet Autonomous Systems (AS) relies on selfish routing strategies that prevent to achieve optimums packet delivery. A more collaborative approach can support QoS intra-AS routing, reducing the global costs (or delay or congestion) of point to point communications.

We are investigating algorithms to compute shortest paths on distributed graphs. The adoption of Secure Mutli-party Computation can support the participant to collaborate without disclosing the sensitive information of their internal networks (e.g. internal network congestion or infrastructure costs).

A main adoption barrier is the commercial nature of AS relations. ASs are competitive parties and selfish, e.g. they are interested in reducing their internal network congestion instead of achieving the common goal. UoA started to apply the existing frameworks to model the intra-AS routing scenario in a game theoretic way, in order to study the inefficiency of noncooperative route traffic through a congested network.