【摘要】：網(wǎng)絡(luò)流量分類技術(shù)在增強(qiáng)網(wǎng)絡(luò)可控性以及加強(qiáng)網(wǎng)絡(luò)管理方面都發(fā)揮著重要的作用。隨著網(wǎng)絡(luò)應(yīng)用的層出不窮,對實時、準(zhǔn)確的流量分類技術(shù)提出了更高的要求,使得近年來研究者大量引入機(jī)器學(xué)習(xí)領(lǐng)域知識來處理流量分類問題,取得了較好的分類效果。但是,特征提取作為機(jī)器學(xué)習(xí)分類算法中一個重要環(huán)節(jié),在處理大數(shù)據(jù)流量時因其計算復(fù)雜度較高、耗時過長已成為制約機(jī)器學(xué)習(xí)算法應(yīng)用于實時流量分類的主要瓶頸。 近年來,GPU硬件體系結(jié)構(gòu)的快速發(fā)展使其浮點(diǎn)運(yùn)算和并行計算能力遠(yuǎn)遠(yuǎn)超過了CPU,在大規(guī)模并行處理和科學(xué)計算等方面取得了廣泛應(yīng)用。特別是NVIDIA公司CUDA編程模型的推出,提供了豐富的API函數(shù),使其能夠更好地發(fā)揮GPU并行處理能力。 本文首先介紹了GPU在體系結(jié)構(gòu)和編程方式上與傳統(tǒng)CPU的不同。其次,對串行特征提取算法執(zhí)行流程進(jìn)行了介紹,并從串行算法每部分的計算任務(wù)大小和特點(diǎn)入手對算法的可并行性進(jìn)行了逐一地分析。在此基礎(chǔ)上,采用CUDA編程語言設(shè)計實現(xiàn)了并行特征提取算法,并利用流化技術(shù)和GPU異構(gòu)執(zhí)行的特點(diǎn)對并行算法進(jìn)行了優(yōu)化。最后,通過實驗對本文提出的并行算法及其優(yōu)化方案在LINUX平臺下進(jìn)行了測試和驗證。實驗結(jié)果表明,在進(jìn)行大數(shù)據(jù)量的網(wǎng)絡(luò)流特征提取時,優(yōu)化后的并行算法相比于串行算法可以達(dá)到2倍以上的加速比,取得了顯著的性能優(yōu)勢。
[Abstract]:Network traffic classification plays an important role in enhancing network controllability and network management. With the continuous emergence of network applications, higher requirements for real-time and accurate traffic classification techniques have been put forward. In recent years, researchers have introduced a large number of machine learning domain knowledge to deal with traffic classification problems, and achieved better classification results. However, as an important part of machine learning classification algorithm, feature extraction has become the main bottleneck in the application of machine learning algorithm in real-time traffic classification because of its high computational complexity and long time consuming. In recent years, with the rapid development of GPU hardware architecture, its floating-point computing and parallel computing capabilities have far exceeded the extensive application of CPU, in large-scale parallel processing and scientific computing. Especially, the introduction of CUDA programming model of NVIDIA Company provides abundant API functions, which makes it better exert the ability of GPU parallel processing. This paper first introduces the differences between GPU and traditional CPU in architecture and programming. Secondly, the execution flow of the serial feature extraction algorithm is introduced, and the parallelism of the algorithm is analyzed one by one from the calculation task size and characteristics of each part of the serial algorithm. On this basis, the parallel feature extraction algorithm is designed and implemented by using CUDA programming language, and the parallel algorithm is optimized by using the fluidization technology and the characteristics of GPU heterogeneous execution. Finally, the proposed parallel algorithm and its optimization scheme are tested and verified on LINUX platform through experiments. The experimental results show that the optimized parallel algorithm can achieve a speedup ratio of more than 2 times compared with the serial algorithm, and obtain a significant performance advantage.
【學(xué)位授予單位】：內(nèi)蒙古大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2014
【分類號】：TP181;TP393.06

【參考文獻(xiàn)】

相關(guān)期刊論文 前3條

1 徐鵬;劉瓊;林森;;改進(jìn)的對等網(wǎng)絡(luò)流量傳輸層識別方法[J];計算機(jī)研究與發(fā)展;2008年05期

2 徐鵬;劉瓊;林森;;基于支持向量機(jī)的Internet流量分類研究[J];計算機(jī)研究與發(fā)展;2009年03期

3 熊剛;孟姣;曹自剛;王勇;郭莉;方濱興;;網(wǎng)絡(luò)流量分類研究進(jìn)展與展望[J];集成技術(shù);2012年01期

，

本文編號：2421942