Research On Dynamic Resource Adaption For Smart And Cooperative Network Architecture

Adapting service demand to network resource has always been a challenging task for Internet architecture design since the current Internet was built. Resource adaption limits unfold in the traditional Internet system, based on which the improvement solutions did not gain satisfactory outcomes. In recent years, service resource oriented future Internet architecture solutions have been studied and proposed worldwide. The Smart and Cooperative Network Architecture, as a service resource oriented Internet architecture solution, introduces the Resource Adaption Layer (RAL). RAL introduces the concept of Family, depends on the family game theoretic strategy to adapt service demand to network resource, and achieves smart content delivery and efficient network resource utilization. This thesis focuses on resource adaption mechanisms including family game strategy based traffic scheduling, content delivery, and cooperation based multipath transmission control. The contributions of this thesis are as follows:Firstly, in the aspect of research on cooperation between service and network layers, this thesis investigates the interaction between application layer selfish routing and network layer traffic engineering, and builds two dynamic non-cooperative game models to describe the selfish underlay-topology-aware behavior of application layer and selfish overlay-traffic-monitoring behavior of network layer, respectively. Unilateral improvement of overlay or underlay gains less profit than the non-cooperation case. Accordingly, a synergistic cost-saving game model is built to deliver content across local networks and optimize network cost for traffic transit.Secondly, in the aspect of research on RAL’s family game theoretic strategy, the game strategy is intelligent because the family can adjust to service demands and network resource availability; the game strategy is reasonable because all the participants gain the optimal profits at the Nash Equilibrium Point and do not gain higher profits with other strategy. This thesis proposes an on-demand-designed cooperative queue model in a family. Based on this queue model, both the delay-sensitive service and loss-sensitive service obtain optimal quality of service (QoS). Meanwhile, the nodes in the network keep load balancing.Thirdly, in the aspect of inter-family game strategy, this thesis proposes four game models to analyze the interdependent relationships between families based on content delivery. Resource adaption for Smart and Cooperative Network is supposed to have the scalability which means cross-familgy resource adaption can be done by cooperation between families. To deal with the cross-family content delivery issue, families can participate in the game to decide their optimal content caching and sharing strategy. The advantage of inter-family cooperation is validated by evaluating the performance of content delivery cost and price of anarchy for the four game models.Lastly, in the aspect of research on transmission control, this thesis proposes a fair and efficient multipath cooperation based control mechanism to satisfy QoS. Content-oriented networks are supposed to support high quality service such as multimedia using multipath transmission. By estimating the path quality for multipath transmission control, multipath flow transmission can be efficient in its throughput and be fair to single-path flows that exist on the same bottlenecks. This is a fair bandwidth sharing mechanism and satisfies quality of transmission and quality of user experience.This thesis focuses on the Smart and Cooperative Network’s resource adaption mechanism including game theoretic strategy and smart control, studies the necessity of RAL, inter-and intra-family game theoretic strategy and fair network resource adaption for intelligent transmission control, and makes contributions to the theoretical system of RAL for Smart and Cooperative Network Architecture. Reference investigation, theoretical model, algorithm design, mathematical deduction and performance evaluation are used as the research methods in this thesis to achieve the research targets.