The Research Of Forwarding Strategy Based On Content And Port State In Named Data Networking

With the continuous emergence of new application in the Internet and the grow-ing demand for network services,the traditional TCP/IP-baesd Internet architecture is showing a variety of inadaptability in mobile,scalability aspects,especially in the emerging patterns of distribution networks expanded by sustained growth in digital me-dia.In order to solve the inadaptability between the current Internet architecture and new applications,domestic and foreign research scholars have put forward the idea of redesigning the future Internet Architecture.As an important member of the future Internet architecture system,NDN(Named Data Networking)has attracted much attention due to its complete routing protocol framework and hierarchical naming rules.In NDN,the network nodes not only pro-vide forwarding capabilities,but also the ability to store.The cache capacity can not only avoid the repeated transmission of data in the network,but also construct an inher-ent multi-path characteristic in data transmission.The research of forwarding strategy,which can make full use of the inherent multi-path characteristics to achieve efficient data forwarding,is of great significance.The main work of this dissertation is as follows:1.Aiming at the problem of out-of-order problem in the current NDN forwarding strategy,this paper firstly proposes a forwarding strategy based on MDP(Markov De-cision Processes).In this method,the Markov decision process model is established in a single node,and then the optimal forwarding strategy is obtained by solving the MDP problem.In the MDP model,as the setting for the system state,the number of different flow in port is used to realise joint aware of port status and content distribution.2.As the lack of joint aware of port status and content distribution in current NDN forwarding strategy,the forwarding strategy based on decision tree is proposed.The whole forwarding process is divided into off-line and on-line.The information obtained in on-line forwarding process is used for off-line learning,and the results of off-line learning are contributed to the better forwarding in on-line process.In addition,the traditional decision tree algorithm is modified to meet the needs of forwarding strategy.3.In this dissertation,two kinds of forwarding strategies have been deployed in ndnSIM2.0,and we evaluate the performance of these two strategies.The experiment results show:1,The forwarding strategy based on MDP is superior to BEST-ROUTE strategy and RANDOM strategy in delay,and distinguish the importance between dif-ferent streams;2,Compared with six forwarding strategies such as OMP-IF,the for-warding strategy based on decision tree has better performance in delay.