Research On Resource Pricing And Dynamic Adaptation In Smart Identifier Network

With the development of Internet applications and continuous growth of network traffic,traditional network operators face severe challenges in resource revenue and resource utilization.On one hand,traditional resource pricing mechanisms do not take into account fine-grained utilization characteristics of network resources such as temporal and spacial heterogeneity.As a result,network operators can not efficiently utilize resource and obtain considerable profit growth,gradually becoming a "dumb pipe".On the other hand,traditional network adopts relative static and rigid resource adaptation mechanisms,which are not effective in adapting network resources dynamically and efficiently,leading to high operating costs and high expansion pressure.Recently,the introduction of new network technologies and architectures such as software defined network,network function virtualization and smart identifier networks makes flexible resource pricing and dynamic resource adaptation possible.Therefore,based on smart identifier networks,the dissertation focuses on the following two questions:1)How to enable network to flexibly price and adapt resources by taking into accounts fine-grained utilization characteristics of network resources?2)How to design effective resource pricing and adaptation approach that fully exploits fine-grained utilization characteristics of network resources and promotes both resource utilization and revenue?The main contributions of the dissertation are:(1)For the first question,the dissertation proposed a framework for customizing network families that are capable of flexible resource pricing and dynamic resource adaptation.The framework enables network to select proper resource pricing and adaptation mechanisms according to the resource requirements of different network scenarios.The dissertation analyzed the design requirements of the framework,introduced key modules and the overall customization process that includes family function customization,pricing mechanism customization and adaptation mechanism customization,and illustrated the lifecycle management procedure.Based on the family customization process,a mathematical model is formulated to convert the customization requirements to resource orchestration strategy and a genetic algorithm is designed to obtain optimal solution in polynomial time.Simulation results show that the framework is reasonable and effective to customize familes with different resource utilization requirements while minimizing network overhead and promoting resource utilization.(2)For the second question,the dissertation proposed an approach to flexibly price and dynamically adapt resources based on time-dependent congestion level of different time slots.The approach sets prices of different time slots differently by taking into account the variation of congestion level of different time slots and conducts price-aware traffic scheduling to efficiently promote resource utilization and quality of experience.The dissertation proposed the main idea of the approach,which is realized by pricing module,traffic measurement module,user proxy and network proxy,introduced the process of pricing and resource adaptation that is realized by the above mentioned modules.By introducing demand queues in user proxy and network proxy,respectively and designing a scheduling algorithm,the traffic scheduling strategy is achieved and executed.Simulation results show that the proposed approach is effective in reducing network overhead,promoting resource utilization and users' experience.(3)For the second question,the dissertation proposed an approach to flexibly price and dynamically adapt resources based on location-dependent congestion level of different locations.The approach sets prices of different sites differently by taking into account the variation of congestion level of different CDN sites of the network and conducts price-aware request mapping,helping the network promote profits and efficiency of congestion management.The dissertation proposed main idea of the approach,introduced the pricing process that takes into account the congestion level of different sites,formulated a mathematical model that takes prices as input and calculates the request mapping rules.Simulation results show that the proposed approach provides better delay performance,reduces network overhead and helps promoting profits by flexible pricing.