Research On Cache Management And Congestion Control In Information Centric Networking

With the development and popularization of information technology,the numbers of network users and terminal devices have shown an explosive growth,and the application scope and scale of the Internet has far exceeded its original design.Users’demands for the Internet have also changed from communication-oriented to contentoriented,and content distribution and acquisition have become the core functions and goals of the Internet.Users are no longer concerned about the source and storage location of content,but the content data itself and how to obtain the content conveniently,quickly and safely.It is difficult for traditional TCP/IP networks to provide efficient and reliable support for content distribution services.In this context,Information Centric Networking(ICN),as a revolutionary future network architecture,has received extensive attentions from both academia and industry due to its good scalability and natural adaptability to content distribution services.The ubiquitous in-network cache is one of the core features of ICN.By deploying cache resources closer to the network edge,users’ content acquisition experience can be significantly improved.However,considering the constraints of economic cost and utilization of resources,the cache spaces in ICN are relatively limited.How to make optimal cache allocation between different content providers(CPs)and different network nodes,and how to make optimal content caching decisions in limited cache resources to improve the revenue of internet service provider(ISP)and the content acquisition experience of users,are both urgent problems to be solved.In addition,the content acquisition in ICN is driven by user,which naturally supports multi-source and multi-path transmission,as well as adaptive request forwarding.The new dynamic transmission model not only facilitates content acquisition,but also brings many new challenges to congestion control in ICN.Therefore,this thesis focuses on the two key issues of cache management and congestion control in ICN,and conducts researches in the following aspects.First,for the problem of cache allocation among multiple CPs in the scenario of information asymmetry in ICN,this thesis proposes a contract theory based allocation scheme with the goal of maximizing the economic revenue of ISP.This thesis firstly models the cache allocation problem as a commodity selling problem in a monopoly market,in which CPs are classified into different types based on their own popularity,and ISP is the monopoly that divide the cache space into blocks of different sizes and set the optimal prices.On this basis,this thesis expounds the construction process and feasibility analysis of the ISP cache allocation contract,and gives the optimal pricing strategy based on the important attributes of the feasible contract.Then the optimal cache sizes allocation is formulated as a global optimization problem,and a solution algorithm based on dynamic programming is proposed.The simulation results show that the scheme can effectively improve the economic revenue of ISP,which verify the economy and effectiveness of the scheme.Secondly,aiming at the problem that small CPs cannot obtain reliable cache services due to the unequal competition among multiple CPs in ICN,this thesis proposes a service guarantee scheme for small CPs based on dedicated cache spaces allocation.Dedicated cache spaces are allocated for small CPs on some core nodes by paying and renting,and core nodes provide dedicated cache services for different edge nodes,which can effectively improve users’ content acquisition experience and the overall cache revenue.This thesis first proposes a core node selection and pairing scheme based on topology information,and designs a core node selection algorithm based on topology importance and a node pairing algorithm based on matching theory.Subsequently,this thesis jointly considers the problem of cache space allocation and cache decision-making on core nodes,and models it as an integer programming problem with the goal of maximizing the overall cache revenue.Then a heuristic algorithm based on linear relaxation and greedy strategy is designed.The simulation results show that the scheme can effectively improve the overall cache revenue and users’ content acquisition experience of small CPs,which verify the feasibility and efficiency of the scheme.Finally,aiming at the key problems of inaccurate detection,untimely feedback and inefficient window adjustment in ICN congestion control,this thesis proposes an intelligent congestion control scheme based on information feedback.Specifically,this thesis first proposes a congestion detection and feedback mechanism at intermediate nodes.The local congestion information of each node is detected by monitoring the buffer occupancy and the number of PIT entries corresponding to different faces.Then the congestion information is filled into the new fields of data packet for feedback to edge users,which ensures the accuracy and timeliness of congestion detection.Secondly,this thesis proposes a window adjustment algorithm based on deep reinforcement learning,which maintains the corresponding link state information for each path and adjusts the corresponding congestion window size through the DRL agent.The optimal control strategy does not depend on any model or prior knowledge,and it also enhances the adaptability to dynamic changes of link states while achieving efficient control.The simulation results show that the scheme can significantly improve the transmission performance,verifying its accuracy and superiority.