【摘要】：隨著信息時(shí)代不斷深入發(fā)展,人們對(duì)于互聯(lián)網(wǎng)的訪問量和需求的增加使得現(xiàn)有的基于TCP/IP的互聯(lián)網(wǎng)絡(luò)逐漸暴露出許多問題以及對(duì)現(xiàn)有互聯(lián)網(wǎng)的不適應(yīng)。目前互聯(lián)網(wǎng)絡(luò)中主要暴露出的問題包括:拓展性差、靈活性差和不安全性等。國內(nèi)外主要有兩種解決思路:一種是不改變互聯(lián)網(wǎng)IP主體地位的思路,不改變主機(jī)到主機(jī)通過建立連接通信的方式,比如對(duì)于IP的拓展方法IPv6;另一種是改變IP網(wǎng)絡(luò)體系結(jié)構(gòu)的思路如NDN(Named Data Network,命名數(shù)據(jù)網(wǎng)絡(luò)),本文主要討論的正是命名數(shù)據(jù)網(wǎng)絡(luò)中的轉(zhuǎn)發(fā)引擎算法的設(shè)計(jì)問題,具體內(nèi)容如下:1、轉(zhuǎn)發(fā)引擎算法結(jié)構(gòu)的基礎(chǔ)設(shè)計(jì)與實(shí)現(xiàn):包括對(duì)于轉(zhuǎn)發(fā)引擎結(jié)構(gòu)的研究,并據(jù)此選擇PatriciaTrie樹結(jié)構(gòu),介紹了PatriciaTrie樹結(jié)構(gòu)的相關(guān)原理,設(shè)計(jì)基于該結(jié)構(gòu)的基本方法;同時(shí)分別列出按字節(jié)劃分、按二進(jìn)制位劃分和按部件劃分的設(shè)計(jì)模型,并詳細(xì)闡述前兩種模型的基本設(shè)計(jì)方法和實(shí)現(xiàn)過程。2、轉(zhuǎn)發(fā)引擎算法結(jié)構(gòu)的優(yōu)化技術(shù)的研究:進(jìn)一步分析字節(jié)劃分模型和二進(jìn)制位劃分模型設(shè)計(jì)中需要面對(duì)的問題以及其中的不足,對(duì)基本的實(shí)現(xiàn)結(jié)構(gòu)以及算法進(jìn)行優(yōu)化,同時(shí)研究多核并行方法在其中的實(shí)現(xiàn),并闡述優(yōu)化后設(shè)計(jì)的實(shí)現(xiàn)過程。3、對(duì)轉(zhuǎn)發(fā)引擎算法的實(shí)驗(yàn)分析:針對(duì)本文中提出的轉(zhuǎn)發(fā)引擎的基礎(chǔ)結(jié)構(gòu)和優(yōu)化后的結(jié)構(gòu),從功能和性能兩個(gè)方面進(jìn)行實(shí)驗(yàn)評(píng)測(cè)。在功能實(shí)驗(yàn)部分,主要針對(duì)結(jié)構(gòu)設(shè)計(jì)的正確性和完整性的驗(yàn)證,對(duì)程序中實(shí)現(xiàn)對(duì)于功能模塊插入、查找、刪除和更新操作正確性的驗(yàn)證。在性能實(shí)驗(yàn)部分,分為兩個(gè)部分:第一對(duì)于內(nèi)存使用量方面,針對(duì)本文中的幾種結(jié)構(gòu)設(shè)計(jì)方案,采用大規(guī)模測(cè)試集進(jìn)行測(cè)試,并得出各種不同的設(shè)計(jì)結(jié)構(gòu)對(duì)于內(nèi)存的消耗,并比較各種方法之間的特點(diǎn);第二對(duì)于時(shí)間效率方面,針對(duì)轉(zhuǎn)發(fā)引擎中插入和查找兩個(gè)主要功能模塊,集中進(jìn)行時(shí)間效率的測(cè)試,分析比較不同結(jié)構(gòu)方案不同功能模塊之間的時(shí)間效率,并比較不同設(shè)計(jì)之間的特點(diǎn)。綜合幾種結(jié)構(gòu)設(shè)計(jì)的多個(gè)方面比較,分析得出轉(zhuǎn)發(fā)引擎中的最優(yōu)的設(shè)計(jì)方法,為進(jìn)一步開展高速命名包轉(zhuǎn)發(fā)引擎的研究工作做鋪墊,從而推進(jìn)整個(gè)NDN下一代互聯(lián)網(wǎng)絡(luò)體系結(jié)構(gòu)的發(fā)展。
[Abstract]:With the continuous development of the information age, the increase of people's access to and demand for the Internet makes the existing Internet based on TCP/IP gradually exposed a lot of problems and the current Internet does not adapt. At present, the main problems exposed in the Internet include: poor expansibility, poor flexibility and insecurity. There are two main solutions at home and abroad: one is not to change the main status of the Internet IP thinking, do not change the host to host through the establishment of connection communication, such as the IP extension method IPv6; The other is to change the idea of IP network architecture such as NDN (Named Data Network, named data network. This paper mainly discusses the design of forwarding engine algorithm in named data network. The main contents are as follows: 1. The basic design and implementation of the forwarding engine algorithm structure: including the research of the forwarding engine structure, and select the PatriciaTrie tree structure, introduce the related principle of the PatriciaTrie tree structure, design the basic method based on this structure; At the same time, the design models divided by byte, binary and component are listed respectively, and the basic design method and realization process of the first two models are described in detail. Research on optimization technology of forwarding engine algorithm structure: further analyze the problems and shortcomings in the design of byte partition model and binary bit partition model, and optimize the basic implementation structure and algorithm. At the same time, the realization of multi-core parallel method is studied, and the implementation process of optimized design is described. 3. The experimental analysis of forwarding engine algorithm: aiming at the basic structure and optimized structure of forwarding engine proposed in this paper, From the function and performance of the two aspects of the experimental evaluation. In the function experiment part, the correctness and integrity of the structure design is verified, and the correctness of the operation of inserting, searching, deleting and updating the function module is verified in the program. In the performance experiment part, it is divided into two parts: first, for memory usage, aiming at several structural design schemes in this paper, we use large-scale test set to test, and get the memory consumption of different design structures. The characteristics of different methods are compared. Second, regarding to the time efficiency, aiming at inserting and searching two main function modules in the forwarding engine, the time efficiency is tested centrally, and the time efficiency between different function modules of different structure schemes is analyzed and compared. The characteristics of different designs are compared. Based on the comparison of several structural design aspects, the optimal design method of forwarding engine is obtained, which paves the way for further research on high-speed named packet forwarding engine. So as to promote the development of the whole NDN next-generation Internet architecture.
【學(xué)位授予單位】：北京理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2014
【分類號(hào)】：TP393.0

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 吳建平;劉瑩;吳茜;;新一代互聯(lián)網(wǎng)體系結(jié)構(gòu)理論研究進(jìn)展[J];中國科學(xué)(E輯:信息科學(xué));2008年10期

，

本文編號(hào)：2308136