GameTE: A Game-Theoretic Distributed Traffic Engineering in Trustless Multi-Domain SDN

With growing network service demands, rational and efficient multi-domain resource allocation is paramount. Research aims to develop intelligent Traffic Engineering (TE) algorithms that can dynamically allocate resources, adapt to changing conditions, and meet user needs. TE algorithms based on Software-Defined Networking (SDN) have proven effective for this goal by leveraging the centralized control plane and programmable data plane of SDN. This enables flexible and dynamic optimization of routing and resource allocation across multiple domains to meet traffic demands. However, domains operated by different service providers may exhibit non-cooperative behavior due to conflicts of interest and competition. Some domains may act selfishly by hiding bandwidth or exaggerating inter-domain requests to reserve more resources for themselves. This complicates TE design as algorithms can no longer assume universal cooperation in multi-domain networks. Game theory provides a framework to model competition between domains through behaviors like request forwarding, dropping and study cooperation strategies. This paper presents GameTE, a game-theoretic distributed TE algorithm for multi-domain SDN environments without trusted relationships between domains. By incorporating incentives and punishments, our algorithm suppresses selfish behaviors and promotes efficient resource utilization. Evaluation results demonstrate that GameTE is effective in curbing deception, enhancing resource sharing between domains, and improving overall network performance compared to baseline schemes.