Title: A Multipath Load Sharing Framework Using Game Theory

Nowadays, the Internet plays an important role in peoples' daily lives as well as their work, thus its resiliency is vital within its ecosystem. However, the uncoordinated patterns of inter-domain routing as well as the significant growth of the network and the traffic in Internet challenge the resiliency of Internet. Therefore, it is becoming urgent to find solutions to mitigate those issues. This dissertation studies how to improve the resiliency of Internet by utilizing multipath load sharing and enforcing the coordination within inter-domain routing with Game Theory. It first analyzes the resiliency issues of current Internet routing in details. After stating the goal to improve the routing resiliency between two independent distant edge nodes in a large topology, it reviews the existing and emerging techniques that may be applied to improve the routing resiliency. Then, a load sharing framework, developed with game theory, is presented to improve the resiliency of inter-domain routing through a cardinal potential game. In addition, a vectorized model, based on a vector space and game theory, is proposed to deepen the depiction of inter-domain routing. Meanwhile, a set of novel refinement values is defined to assist the multipath evaluation for the new model, which brings a generalized multipath load sharing framework. Multi-criteria simulations on real instances show that significantly higher routing resiliency can be achieved through the universal multipath load sharing framework.