Research On Resource Naming And Distribution For Future Networks

With the wide spread of Internet applications, the current Internet system reveals many shortcomings, which seriously affect the sustainable development of the Internet. Future Internet Architecture (FIA), with great significance, is designed to fundamentally solve these drawbacks. Currently, the worldwide researches on FIA are still in the starting stage. There are a number of key mechanisms and solutions to be explored and proposed. Among them, resource naming and distribution are the core researches on future network, and are desired to be focused on and promote the related standards. To this end, this thesis focuses on the key technologies of resources naming and distribution in future network and make a series of detailed studies. The main work includes two aspects:resource naming and resource distribution, which are further divided into four specific research contents:(1) resource naming scheme;(2) resource name lookup mechanism;(3) resource distribution model, and (4) resource cooperative caching strategy, which build an entirely resource naming and distribution system for future network. The main contributions are as follows:Firstly, in terms of resource naming scheme, this thesis systematically studied the naming schemes and designs for future networks. Considering the requirement of fuzzy matching during resource searching, to make up the drawbacks of flat naming and hierarchical naming, this thesis proposed a novel naming scheme named by M-Label, which can support the fuzzy matching in the view of name level. Due to the great flexibility and scalability, M-Label scheme can make network take advantage of these properties, easier and flexible to meet the diverse needs of various users, and achieve efficiently resource distribution. Further, through analyzing of hierarchical naming, flat naming and label-based naming, this thesis also refined the advantages and disadvantages of above naming structures, cleverly presented a novel "Layered-Flat-Label" based resource hybrid naming scheme, named by Hybrid Naming (HN). The HN scheme achieved a compatible naming scheme for a variety of future network systems, which can improve the compatibility of future network, and improve the interoperability among future heterogeneous network systems.Secondly, in terms of name lookup research, this thesis researched the longest prefix match (LPM) principle in routing table, and made in-depth analysis of different implementation methods for LPM in IP routers as well as their advantages and disadvantages. To accommodate the resource-oriented routing lookup, this thesis put forward a Bloom Filter and Trie-based hybrid structure for storing and matching names. This structure reduces the number of entries inserted in the Bloom Filter, thereby reduces the possibility of hash collisions with respective to the traditional Bloom Filter based lookup scheme. On the other hand, this structure can solve too many memory accesses problem due to the length of names being too long. Further, an adaptive optimization mechanism based on prefix popularity was also proposed. Because different resource names with different popularity levels may have different matching frequency, adaptive Bloom Filter name prefix lookup was proposed. This adaptive mechanism can effectively accelerate the lookup rate of matching popular resources names, then, accelerate the overall rate of the router to find a match.Thirdly, in terms of resource distribution modeling, this thesis considered the features of content distribution in future resource-oriented network system, content caching mechanisms as well as node mobility models, and built a complete hit/miss ratio based resource distribution models, which are from two views of one-dimensional and two-dimensional moving scenarios respectively. Through mathematical theory derivation, the analytical expressions were derived for future mobile network environments, such as the average throughput and the average transmission delay. As analysis tools for resource distribution mechanisms, the proposed models will provide great benefits for explore potential in future network system, and promote theory analysis and evaluation on future network performance optimization, and improve the related mechanisms research.Fourthly, in terms of cooperative caching resources, this thesis researched the cooperation among network nodes to improve the resource distribution efficiency. For the static network case, this thesis first studied the graph theory based minimum coverage set theory to solve the problem of resource caching nodes selection optimization. For dynamic mobile scenarios, thesis took the highway mobile vehicle application as an example, focused on the research of social relationship based resource caching mechanisms. By analyzing the different roles of vehicle nodes "partner node" and "courier node", social relations based cooperative caching mechanisms were proposed, which include "partner-based cooperation" and "courier-based cooperation". The proposed cooperative cache mechanisms enable to facilitate the efficient distribution of resources to some extent.In summary, this thesis focused on the resource naming and distribution problems in future network, and made a series of innovative researches following a four-step line of "Naming-Lookup-Distribution-Caching", which built a whole research system. In research methods, this thesis adopted the contents investigation, principles mechanism design, statistical analysis, mathematical derivation, simulation assessment research methods, which jointly contributed to the research results. In the applications, the researches considered the practical application of future network deployment, including:resource naming compatibility with existing Internet, distribution model application guidance, as well as cooperative caching mechanisms used in the future mobile vehicle network and so on. The work of this thesis will contribute to the research and development of future networks.