QoS On-demand Oriented Multi-user Stackelberg Game And Network Resource Pricing Research

With the rapid development of the Industrial Internet and 5G,the network cannot meet the individual needs of multiple users.In the network service market,internet service provider(ISP)has absolute pricing power in network pricing and resource allocation,while users as consumers of network resources have no bargaining power and have to accept network services passively.Based on Software Defined Networking(SDN),this paper proposes a new network architecture named User-enabled Defined Networking(Ue DN),where users are entitled to define the network to a certain extent to meet their own personalized Quality of Service(QoS)requirements.Furthermore,users are given reasonable bargaining power.In the context of the Industrial Internet plus 5G,this business model is extremely attractive to price-sensitive users with high traffic.In Ue DN,users can define the network according to their individual needs and pay on demand;while ISP can meet user needs by customized services and obtain higher profits by differentiated pricing.Based on Ryu,a component-based SDN controller with high programmability,and Mininet,a powerful virtual network platform,we completely implement the prototype of Ue DN,and conduct some functional simulations to verify the feasibility of Ue DN.Considering the scarcity of network resources,it comes naturally that there is a game relationship between ISP and users on pricing and resources.This paper quantifies the goals of the two participants in the game through mathematical modeling,builds Stackelberg Game to get deep insight into the competition,and proves the existence of Nash Equilibrium.By comparing the backward induction which requires complete information,we propose a distributed iterative pricing algorithm based on Stackelberg Game which requires only partial information.In algorithm simulation and analysis section,we verify the speed of dynamic convergence of the algorithm,explain the details of the entire game process from different perspectives,and also explore the impact of various factors on the game equilibrium.Finally,comparing with the existing usage-based pricing,we find that the proposed algorithm based on game theory has better performance in ISP revenue and is more suitable for present auto scaling network.