Research On Modeling And Content Placement In Content Centric Networking

Internet has become one of the most important communication infrastructures in nowadays world and changed the way of our working, studying and living. However, along with the rapid increasing of the network scale and the continued emerging of the multiple requirements, the inherent problems of nowadays Internet architecture have become much harder to avoid, such as scalability, mobility, QoS, and energy saving. To overcome these problems fundamentally, there emerges a view of designing the future network architectures in a clean-slate way. They all want to break through the current Internet and design the data-oriented network architectures.Content Centric Networking (CCN) or Named Data Networking (NDN) is one of the most important future networks of the data-oriented network architectures. Based on the content-centric, CCN/NDN supports the interest-data packet communication model, which changes the Internet fundamentally by turning the attention from the address or host to the content. Instead of the pipe transferring the data, CCN/NDN protects the data directly. Thus we can trust the data other than the host. Furthermore, CCN/NDN inputs the storage in router to improve the transfer efficiency by the in-network caching.In this thesis we make our effort to CCN/NDN modeling and content placement algorithms. Under the assumptions that the network request distribution, caching policy and routing scheme are given, we model the data transfer to get the hit probabilities and hit rates, and the average content delivery time in CCN/NDN. The main innovations proposed about modeling in this paper are as follows:(1) In the linear and binary tree topologies employing the Least Recently Used (LRU) cache policy, both of the two cases of miss at a router are considered:i) when the arrival time interval between the two requests for the same content chunk is greater than the Residual Virtual Round Trip Time (RVRTT), the sum of different chunks arrived is larger than the cache size; ii) when the arrival time interval between the two requests for the same content chunk is less than the RVRTT. Then we deduce the closed-form expressions of miss rate and the Virtual Round Trip Time (VRTT) based on the big queen theory and the law of great numbers. Finally the relationship between the performance and resources is analyzed.(2) For the general topology of CCN/NDN, the Multi-Cache with Aggregation Approximation (MCAA) algorithm is proposed to get the results of content miss rate and average delivery time at each router. From the data transfer feature of CCN/NDN, the aggregation of Pending Interest Table (PIT) is considered. Modeling the request rates, miss rates, forwarding rates with aggregation and the average delivery time by expressions, the MCAA algorithm sets the miss rate as zero initially and calculates the expressions above step by step and iterate them until the error is smaller than the predefined precision threshold. Thus we get the approximate content miss rate and average delivery time at each router in CCN/NDN.The research about content placement problem can be separated as two classes of centralized and distributed content placement. The content can be placed at the optimal place globally as the global information is given in centralized. While the distributed content placement just get the second-best solution with the local information. As CCN/NDN is a distributed architecture, this paper studies the distributed content placement problem. The main problems to be solved in CCN/NDN content placement are following:whether the content is placed at the node (or the decision policy), when the cached content to be removed (or the cache replacement policy), and how many contents should be placed at the node (or the cache size allocation policy). In this paper, we study the content placement based on the optimal objections about delay and energy efficiency respectively, and the main contributions are following:(3) In order to minimize the delay of users obtaining the contents, we design a novel cooperation cache decision policy based on the Sojourn Time (ST), which is defined as the time from the content to be cached to being removed at the node. We analysis the advantages of ST in theory by deducing the closed expression of ST employing LRU cache policy. By comparing with the tradition scheme, the numerical results show that the new scheme increases the hit rate and decreases the total hops through increasing the caching content diversity.(4) In order to optimize the energy efficiency, we derive expressions to tradeoff the caching energy and the transport energy based on the energy optimization model of CCN with in-network caching, and then we design a new Energy-efficient caching scheme based on Virtual round trip time (EV) in CCN. The EV policy designs two criterions to solve the three problems in content placement:i) deciding if the cache energy of the new content is smaller than the transport energy; ii) deciding if the energy saving of the new candidate is larger than the cached contents. From the simulation results, we know the EV scheme is better than the LRU and popularity based cache policies on the network average energy consumption, and its average hops also much better than LRU policy.