Research On Information Center Network Cache Optimization Mechanism

With the development of the Internet,more and more devices are connected to the Internet which cause network congestion.And the bottleneck of the current IP network architecture has gradually emerged.Therefore,the information center network is proposed as a new type of Internet architecture.In this architecture,the main communication mode is no longer host-tohost,but content-to-content.Due to no attention on the source of the content,and content consumers can get content by content name or an identifier,so the information can be cached by the router which increases the network efficiency.It relieves the excessive pressure of the backbone network.However,the movement of content producers and cache pollution detection have not been well solved.These problems are also urgent problems to be solved in the current information center network.First,in the information center network,content delivery is driven by content consumer,when the content producer moves,content consumers can not sense movement,so content will still pass in the direction which before moved.There may cause transmission delay of the content increased or interruption,resulting in an increase in the network cache redundancy.In this paper,a cache routing optimization strategy based on redirection is proposed for this problem.This strategy notifies the original access location by sending a control packet after the content producer moves,and the content consumer can perceive the mobile through the feedback mechanism,and the content consumer can redirect to the new location to obtain the resource.At the same time,the strategy can cache location change information in the route along the way,so that the network has good mobile sensing capabilities.This strategy has good path scaling characteristics,which can reduce cache redundancy and use the data cache efficiently meanwhile maintain a good interest package convergence ability.The comparison experiments show that the routing strategy can reduce cache redundancy and transmission delay while ensuring efficient content distribution.Secondly,Cache is a very important part of the information center network.Therefore,cache pollution attacks are one of the most devastating attacks in information center networks.In the research of cache pollution detection,most existing researches to detect cache pollution via manual threshold setting which has weak applicability.For different network environments,thresholds may need to be reset.The accuracy is also related to the threshold setting.Therefore this paper proposes a cache pollution detection algorithm based on Gradient Boost Decision Tree(GBDT),which can learn the detection method of cache pollution through model learning.In characteristic selection,the algorithm adopts node state and path information as the input of the model.The node state feature can reflect the distribution of the request,and the path information feature represents the requested path.These two characteristics makes the method improved in accuracy.In this paper,the improvement of detection accuracy is proved by comparative experiments.