Research On In-Network Caching Schemes For Content Centric Networking

In parallel to the Internet’s growth, an unprecedented number of innova-tions, in both the applications and services running on top of it have emerged, such as the streaming media, cloud computing, Internet of things, and social networking services. The most applications in the Internet are to fetch the mass contents drived by receivers instead of the end-to-end communication services. However, the content cannot be efficiently distributed by the overlay content delivery technology such as CDN and P2P because of the fact that they try to match the content centric problem with an infrastructure which is built on top of a host-to-host communication model. All above lead that amount of duplicated contents are transmitted on the links, which reduces the transmission efficiency seriously.Information-Centric Networking (ICN) has emerged as a promising can-didate for the architecture of the Future Internet to resolve the problems of the scalability of data transmission, the mobility, and the security. As an innova-tive network paradigm integrating content delivery as a native network feature, ICN brings the change from "Where" to "What", which aims to supply the content-oriented functions instead of focusing on the position of the informa-tion.Among these proposals, Content Centric Networking (CCN) is one the most promising:in CCN, CCN communication is driven by the consumers of data, and content is sent in reply to interest packets addressing named data chunks, which gets cached along the way back to the request originator. C-CN paradigm can reduce the redundant data and bandwidth consumption by caching and utilizing the duplicate content on the middle routers.CCN takes content as a primitive decoupling location from identity, secu- rity and access, and retrieving content by name. Using new approaches to rout-ing named content, derived heavily from IP, we can simultaneously achieve scalability, security and performance. One of the salient CCN features is in-network caching, with potentially every network element router caching all content fragments that traverse it.The transparent, ubiquitous in-network caching is one of the key aspect-s of CCN, which speeds up content distribution and improve network re-source utilization. Although caching theory and techniques to optimize the caching system have already been extensively studied, new features of CC-N caching-transparency, ubiquity and fine-granularity-have made traditional caching theories, models and optimization techniques developed for hierarchi-cal Web caching and CDN caching systems unable to be directly and seamless-ly ported to CCN caches.In this thesis, we aim to resolve two issues to improve the scalability of data transmission. One is the cache decision scheme, which decides which objects to be placed at which nodes. And the other is cache dimensioning, which provides the method to allocate the cache resources. There are three contributions as follows.1) Propose two novel in-network caching schemes respectively based on the node state and the actual content popularity on the node.The in-network caching is one of the key aspects of CCN, which speeds up content distribution and improve network resource utilization. It is valuable to make good use of the cache resource and the bandwidths to solve the location of the respond object along the delivery path.Firstly, a novel in-network caching scheme named BetwRep is "proposed based on a metric including the betweenness centrality and the replacement rate of one node to address the problem-where to cache along the delivery path. The scheme ensures that the more important nodes are chose to cache as the cache can be adjusted be the replacement rate, which avoids the objects are replaced with a high rate at the more important node. Simulation experiment based on ndnSIM demonstrates that the BetwRep caching scheme achieves the lower loading in the source server and less average hops than that of Betw scheme and ALWAYS scheme.Then, based on the node state and the actual content popularity on the n-ode, a probabilistic in-network caching scheme has been designed to calculate the probability to cache on every node along the delivery path to realize the op-timal placement of the respond content. The core idea is that the more popular contents are expected to cache at the more important nodes.2) Present a new in-network caching based on grey relational analysis decision (GRAD).CCN performs request aggregation. Nodes toward the core on the topolo-gy may have received less interest packets because of the request aggregation while it is opposite in the queuing network, which results in that the node’s importance. It cannot be directly measured by the number of the interest pack-ets receiving. The Request Influence Degree (RID) is first time defined in this chapter, distinguish the degree of nodes’importance along the content delivery path by analyzing the interest packets received.Then, combining the replacement rate, hops a request traveling from the users, and the node traffic, four network attribute parameters are considered during the in-network caching algorithm design. Based on these four network parameters, a GRAD based in-network caching algorithm is proposed, which can significantly improve the performance of CCN. And the Analytic Hierarchy Process (AHP) is used to calculate the weights of the above four parameters.3) Address two innovative content store space heterogeneous allocation schemes respectively based on the metric RID across the CCN network and the optimal energy consumption. And analyze the effect on the cache resource allocation from the cache decision policy.Firstly, a newly cache assignment scheme based on the metric RID across the CCN network has been proposed to solve the content store space heteroge-neous allocation problem. Numerical experiments reveal that the new scheme can decrease the routing stretch and the source server load contrasting that of the homogeneous assignment and several graph-related centrality metrics allo-cations.Secondly, we analyze the effect on the cache resource allocation from the cache decision policy by simulations. Finally, a minimum cache resources allocation optimization model based on energy consumption is made, which is solved by the Genetic Algorithm to obtain the optimal allocation scheme of the cache resource.