Research On Resource Coordination For Information-centric Networking

With the rapid growing of the network,especially the widespread of mobile phones,wearable devices and other intelligent devices,people's daily life has been deeply im-pacted by the network.However,the existing network,built decades years ago,still needs to overcome lots of stern challenges,such as,huge Web content delivery and information security protection.To accommodate those challenges,researchers have proposed to build new network architecture to replace it.The information-centric net-work(ICN),one of the most outstanding next generation networks,proposes to use name based routing,dynamic forwarding,etc.The proposed ICN paradigm is still in the early stage of growth which needs to be studied in depth,especially resource coordination and optimization.ICN has shift the host-based network to a content-oriented one because users and applications care about the content not where and how it comes from.Name,worked as identifier,is used to look up and forward by network's intermediate nodes.Users obtain the content by sending a request which carries the name.In-network caching enables ICN to cache content inside the network to satisfy future requests.Dynamic forwarding is also supported by ICN,which can forward requests to next hops dynam-ically based on the name.In addition,ICN also supports content signature,stateful forwarding etc.The new features of ICN not only enhance network's ability but al-so influence the design of the congestion control.Because of dynamic forwarding and in-network caching,traditional congestion control methods can' t be used by ICN without modification.Besides,the existing Explicit Congestion Control based methods can' t take full use of multipath because of the inability of maintaining end-to-end sta-tus.Hop-by-hop based methods also attracts lots of attention,but it's difficult to store huge of forwarding status.From the aspect of computing resources,Service Centric Networking(SCN)and Named Function Networking(NFN)are network models which can provide distributed computing built on ICN.SCN is a more general service cen-tric network where networked software functions are dynamically deployed,replicated and invoked.NFN proposes the name can be composed of the ?-expression which can be executed by the ICN router.However,they did not consider the impact on packet forwarding when occupying computing resources.To make full use of network and computing resources of ICN,in this paper,cor-responding to the limitation above,we propose a gateway-aware congestion control mechanism and distributed computing task scheduling mechanism based on ICN.First,we investigate the model of gateway-aware congestion control.In particular,we studied the congestion control technology based on gateway revenue maximization and use price to control the user's request rate and dynamic forwarding ratio.Because ICN uses gateway to connect lots of Autonomous System,the gateway can be aware of traffic load and congestion level.With the feedback price provided by the gateway,the users adapt their request sending rate to achieve congestion control.Supposing that users are rational and price-sensitive,they need to pay the gateway to use service.The gateway makes reasonable price to maximize revenue based on congestion level.When congestion happened,users reduce the request rate according to the higher price which is sent by the gateway.In addition,the routers use the received price for adjusting forwarding ratio.Second,we study the distributed computing task scheduling problem based on ICN.Compared with centralized computing,distributed computing decomposes com-puting tasks into several small tasks,and assigns them to multiple servers.Distributed computing can push the limit of hardware requirements,reduce computing time and improve computation efficiency.However,redundant computing tasks and storage of computation results need be concerned.In addition,network transmission time also affects efficiency.In this paper,we study the possibility of using ICN hosts to execute computational tasks.Based on in-network caching,the code of computation tasks and the results can be cached in ICN for future usage.Task scheduling can also be achieved simply but efficiently according to dynamic forwarding.