A Research On Cache Mechanism Of Future Network

With the rapid development of information era,traditional host-centric end-to-end communication cannot meet users' needs.Today's Internet is increasingly used for information dissemination rather than pair-wise communication between end hosts.Innovative Content-Centric Network architecture was proposed to handle these problems.In CCN,a file is divided into several chunks and every chunk has its unique name.Information Dissemination and searching are all based on these names.It realized content decoupled from its physical location.In addition,CCN has innetwork caching mechanism.Content routers cache chunks and serve as servers to deliver them to requesting users.This characteristic can effectively reduce server load and avoid network congestion.In this thesis,research is focused on the caching mechanism of CCN.Caching mechanism in CCN is mainly divided into cache decision strategy and cache replacement strategy.In terms of caching decision,a collaborative cache decision strategy based on popularity is proposed in this thesis.Popular contents are preferentially cached in CR nodes.With routers cooperated with each other,the popular contents can be gradually pushed to the downstream.It can reduce the delay that users access to them.Because of the cooperation between routers,there is only one copy of a certain chunk on the transmission path which can effectively reduce cache redundancy.Apart from these,the content searching is based on the caching history trails created by CR nodes along the chunk delivery path.Then requests can be efficiently forwarded.Simulation results show that the algorithm can effectively improve the cache hit ratio and reduce the RTT.In terms of caching replacement,a popularity-driven cache replacement strategy is proposed in this thesis.The popularity of each chunk is calculated based on the Exponential Weighted Moving Average Function which can better adapt to the dynamic network environment.In addition,the position of popularity monitor is changed in this algorithm which can more accurately monitor chunks' popularity.When the cache space is not enough,the chunk with the lowest popularity will be evicted firstly in order to release the cache space for newly arrived data.Simulation results show that,compared with the traditional replacement strategies,it can effectively improve the cache hit ratio,reduce server load and improve system performance.