Research On Optimization Of Access Delay In Content Centric Networks

During recent years, Content Centric Networks (CCN) have been put forward with a keyidea of universal caching, which enables every node cache contents to reduce user’s access delay.However, current CCN still incurs a large delay of content access. To sum up, there are thefollowing four reasons:1) Existing caching strategies cannot adapt to changes of user demand,leading low hit rate of servers;2) The universal caching scheme causes content redundancyamong nodes;3) Existing content routing mechanisms lack of routing to cached content incontrol plane, leading requests to be forwarded to servers with longer delivery paths;4) Existingcontent routing algorithms have no regard to changes of network state, causing a poor delayperformance that users perceived.This dissertation’s studies focus on the optimization of content access delay in CCN. Firstly,with the goal of―making content closest to users‖, the content placement algorithm for replicaservers and the caching policy for routing nodes are proposed from the aspect of―selectingplacement locations‖. Then the key idea―intercepting en-route content requests‖for delayoptimization is established and the shortcut routing mechanism as well as the service nodeselection algorithm is proposed, so as to shorten the―trip‖that requests are forwarded fromproduction to hit the requested content and reduce the content access delay. The research worksin the dissertation are as follows:1) A cooperation-based content placement algorithm for replica servers is proposed.Existing content placement algorithms do not resolve the territorial and dynamiccharacteristics of quests, and lead to low hit rate. For this problem, the delay optimization modelof content placement is created and the concept of content activity is introduced to reflect thedegree of frequency and proximity that the content has been requested, and then acooperation-based content placement algorithm (CCPA) is presented. As simulation results show,the CCPA algorithm improves server hit ratio as well as satisfies the variable demand of clientswith low access delay.2) An adaptive caching policy for CCN nodes is presented.In order to resolve the cache redundancy problem, the dissertation simplifies in-networkcaching as an in-path caching problem and proposes an adaptive caching policy. The ACAalgorithm replaces content with least activity and selects content with higher activity than somecontent in caches to enter. Simulation results show that ACA algorithm reduces the hop thatcontent requests hit.3) A shortcut routing mechanism based on replicas announcement is put forward.Content requests in existing content routing schemes have a long respond path till their hits. For this problem, a shortcut routing mechanism based on replicas announcement is proposed toresolve route construction for cached contents on nodes. In shortcut routing, the validity ofcontents in caches are evaluated, then CCN nodes selectively announce cached contents toreduce cache miss ratio and limit announcement scope to reduce cost. Simulation results showthat the shortcut routing has effectively reduced the average access delay.4) A service node selection algorithm based on ant colony optimization is proposed.To resolve the dynamic routing problem, an Ant Colony Optimization based distributedservice node selection algorithm (ADSS) is proposed. The ADSS algorithm uses path delay,server load and content activity as criterions for service node selection and applies ant colonyoptimization to search for the optimal service node. Simulation results show that as thesimulation getting stable, the ADSS algorithm reduces the average access delay by4%comparedto the shortest path forwarding strategy. Meanwhile, the convergence time of ADSS is relativelyshort.