Research On Caching Methods For Multi-objective Optimization In Information-centric Networks

As a promising future network architecture,Information-Centric Networking(ICN)has attracted widespread attention for its design idea.In ICN,users mainly focus on obtaining information without regard to their relative locations.Thus,ICN provides in-network caching to accelerate content distribution and improve user experience.The core idea is that in-network nodes(e.g.,router)equipped with cache space can store passing contents,so that the subsequent requests can be directly satisfied by the intermediate cache nodes without accessing the original server.Due to the limited cache space,a reasonable and efficient cache placement method plays an essential role in enhancing cache efficiency for ICN.Therefore,it is valuable to study cache placement method in ICN.This thesis firstly introduces the research background and current situation of cache placement methods in ICN.Then,some typical cache placement methods and performance evaluation metrics are described in detail.In addition,by considering multiple optimization objectives,two new cache placement methods are proposed.Finally,the proposed cache placement methods are analyzed based on several performance evaluation metrics.The main research work of this thesis includes:(1)C ontent centrality metric is used to measure the importance of nodes in the network when users acquire content.Different from those centrality metrics in the traditional network,content centrality metric considers not only the location centrality of nodes but also the attribute of content.(2)A probabilistic caching method based on content centrality and delay(ProbCCD)is proposed to improve cache hit and utilization,reduce delay and redundancy.Firstly,by considering cache hit and network delay,content centrality and delay saving metrics are defined respectively.Then,the content placement model taking into account the optimization objectives of both content centrality and delay saving is established.Meanwhile,ProbCCD method is proposed to solve the optimization model.In ProbCCD,each cache node along the delivery path makes a cache decision with a certain probability,which is proportional to content centrality and delay saving.Simulation results show that ProbCCD method can effectively improve cache hit and content caching utilization,and reduce delay and cache redundancy.(3)Aiming at the problems of load imbalance and low cache efficiency in ICN,by jointly considering load balancing,delay and cache hit,a partition-based load balancing caching method(PLBC)is proposed.In PLBC,the partition of the network is introduced firstly and the partition considers both load balancing and delay.Then,based on the partition,local content centrality metric is used to select cache nodes,and the content is collaboratively placed in each partition.Simulation results show that PLBC method can effectively improve cache efficiency and content diversity,achieve load balancing on the whole network,and reduce cache redundancy.