| （英） |
Running network functions as software on generic hardware is a promising technology called as network functions virtualization (NFV).In NFV networks, a network service can be realized as a service chain, which is a sequence of virtual network functions (VNFs).When a new service chain request arrives at the NFV network, a service broker solves a service chaining problem by constructing a special path, called a service path, from the source to destination, which satisfies both the capacity constraints on nodes and links as well as executing the VNFs in order.In this paper, we focus on competitive situations where multiple service brokers aim to select a service path to minimize the service path delay for its own customers.With the help of the existing distributed route selection scheme based on game theory in road networks, we propose competitive service chaining among multiple service brokers in NFV networks.Through simulation experiments, we show the proposed method can achieve low-latency service paths with exponential rate of convergence. |