本文選題：擁塞控制　切入點：速率控制協(xié)議　出處：《寧波大學(xué)》2015年碩士論文

【摘要】：隨著網(wǎng)絡(luò)技術(shù)的快速發(fā)展,互聯(lián)網(wǎng)已經(jīng)成為人們生活不可缺少的一部分。雖然隨著網(wǎng)絡(luò)硬件技術(shù)的發(fā)展網(wǎng)絡(luò)帶寬不斷增大,但是網(wǎng)絡(luò)資源總是有限的,由于網(wǎng)絡(luò)中資源和流量的分布不均勻,網(wǎng)絡(luò)擁塞的發(fā)生就成了互聯(lián)網(wǎng)的固有屬性。為了增加互聯(lián)網(wǎng)的帶寬利用率、提高互聯(lián)網(wǎng)的穩(wěn)定性,需要執(zhí)行擁塞控制。根據(jù)擁塞控制算法運行的位置,可以分成源算法和鏈路算法。源算法只涉及網(wǎng)絡(luò)邊緣設(shè)備,根據(jù)網(wǎng)絡(luò)邊緣設(shè)備獲取的信息調(diào)整數(shù)據(jù)發(fā)送速率,如TCP中的擁塞控制機制就是利用網(wǎng)絡(luò)端主機獲取的信息來預(yù)測網(wǎng)絡(luò)擁塞的發(fā)生;鏈路算法則是通過數(shù)據(jù)包頭收集網(wǎng)絡(luò)設(shè)備的擁塞信息,將此信息寫入數(shù)據(jù)包,并將擁塞信息反饋給端主機來檢測擁塞的發(fā)生,如顯式擁塞控制算法。傳統(tǒng)的擁塞控制機制往往是通過TCP及TCP改進(jìn)協(xié)議來實現(xiàn),隨著互聯(lián)網(wǎng)技術(shù)的發(fā)展,此機制暴露了許多局限性。本文分析目前TCP擁塞控制機制的基本原理,包括慢啟動、擁塞避免、快速回復(fù)和快速重傳機制,以及一些TCP擁塞控制改進(jìn)協(xié)議,如TCP Westwood、TCP Vegas、HSTCP等,發(fā)現(xiàn)TCP的窗口減小方式在高速網(wǎng)絡(luò)上的性能嚴(yán)重下降,且無線網(wǎng)絡(luò)中因信道誤碼率無法充分利用帶寬。針對這些問題,通過對顯式擁塞控制機制分析,本文進(jìn)行了基于RCP(Rate Control Protocol)的精確速率反饋擁塞控制算法研究,并針對無線網(wǎng)絡(luò)提出一種基于帶寬估計的速率反饋擁塞控制方案,最后提出一種基于速率的MPTCP擁塞控制方案。本文的具體的研究內(nèi)容如下。(1)深入分析了RCP協(xié)議的擁塞控制方式,其使用鏈路容量除以前一時刻的速率來估計當(dāng)前流數(shù)量,在短數(shù)據(jù)流較多且數(shù)據(jù)流數(shù)量變化較大的場景下這種估計方法有一定限制,此時RCP協(xié)議會由于其計算通過路由器的數(shù)據(jù)流數(shù)量偏差較大,造成反饋速率計算不準(zhǔn)確。針對這個問題,提出了一種基于RCP協(xié)議的精確速率反饋擁塞控制算法,該算法略去了估計通過路由器的流數(shù)量,就能精確計算出反饋速率,并在NS2仿真平臺上進(jìn)行了各種數(shù)據(jù)流環(huán)境下的仿真實驗,結(jié)果表明該算法和RCP一樣具有較高的網(wǎng)絡(luò)利用率,并且擁有較低的隊列長度和較短的平均流完成時間。因此該算法能夠在較快的速度下反饋精確速率并具有較短的平均流完成時間,能提升網(wǎng)絡(luò)的應(yīng)用性能。(2)分析了無線網(wǎng)絡(luò)中TCP擁塞控制的現(xiàn)狀和限制,在具有高錯誤率的無線環(huán)境下,AMID擁塞控制方案的吞吐量會急劇下降。發(fā)現(xiàn)其主要原因是傳統(tǒng)的TCP擁塞控制機制不能夠區(qū)分?jǐn)?shù)據(jù)包丟失的原因,即是由無線信道造成的還是有網(wǎng)絡(luò)擁塞造成的丟包,因此在實際網(wǎng)絡(luò)并沒有發(fā)生擁塞,但是由于無線信道錯誤造成無碼丟包的情況下,TCP依舊降低其發(fā)送速率�，F(xiàn)有的基于端到端的帶寬估計擁塞控制方案,都有TCP擁塞控制協(xié)議遺留下來的缺點,比如在無線鏈路下性能較低、產(chǎn)生丟包時吞吐量抖動過大、帶寬利用率不高等。本文提出一種基于帶寬估計的速率反饋擁塞控制方案,并設(shè)計一種帶寬估計方法,將基于RCP協(xié)議的精確速率反饋擁塞控制算法應(yīng)用到無線網(wǎng)絡(luò)中,最后根據(jù)NS2仿真實驗分析,可以獲得較高的吞吐量和穩(wěn)定性。(3)對多路徑傳輸控制協(xié)議進(jìn)行了分析,發(fā)現(xiàn)保證多條路徑同時進(jìn)行可靠傳輸十分重要,同時,對MPTCP的擁塞控制機制進(jìn)行討論,用速率反饋代替基于窗口的MPTCP協(xié)議,提出一種基于速率的MPTCP擁塞控制方案,將精確速率反饋擁塞控制協(xié)議應(yīng)用到MPTCP的每條子流的擁塞控制中,充分提高每條子流的吞吐量,并通過分析,理論上該方案能夠快速利用剩余帶寬,擁有較快的收斂速度和較高的吞吐量。
[Abstract]:With the rapid development of network technology, the Internet has become an indispensable part of people's life. With the development of network bandwidth network hardware technology continues to increase, but the cyber source is limited, due to the distribution of resources and traffic in the network is not uniform, the occurrence of network congestion has become the inherent attributes of the Internet. In order to increase the utilization rate of the Internet the bandwidth, improve the stability of the Internet, need to perform congestion control. According to the congestion control algorithm running position, can be divided into source algorithm and link algorithm source. The algorithm only involves the network edge equipment, according to the information data transmission rate adjustment of the network edge equipment acquisition, such as is the use of network host access to the congestion control mechanism of TCP information to predict the occurrence of network congestion; link algorithm congestion information collection network equipment through the data of Baotou, this letter Write information packets, and congestion information feedback to the host to detect congestion, such as the explicit congestion control algorithm. The traditional congestion control mechanisms are often improved by TCP and TCP to implement the agreement, with the development of Internet technology, this system has exposed many limitations. The basic principle of the analysis of the current TCP congestion control the mechanism, including slow start, congestion avoidance, fast recovery and fast retransmit mechanism, and some of the TCP congestion control protocol, such as TCP Westwood, TCP Vegas, HSTCP TCP, found the window decrease performance in high-speed network and wireless network is greatly reduced because of being unable to make use of bandwidth of the channel error rate. To solve these problems, through the analysis of the explicit congestion control mechanism, this paper based on RCP (Rate Control Protocol) feedback congestion control based on the accurate rate, and according to the wireless network is put forward A bandwidth estimation rate feedback based congestion control scheme, finally proposed a congestion control scheme based on MPTCP. The specific contents are as follows. (1) in-depth analysis of RCP congestion control, the use of the link capacity in previous time rate to estimate the current flow quantity, under the estimation method must limit the number of data streams flow more and larger changes in a short scene data, the RCP protocol will be due to the calculation of the number of data streams through the router large deviation, caused by inaccurate rate feedback. Aiming at this problem, this paper proposed a feedback congestion control algorithm accuracy rate based on the RCP protocol, the algorithm the number is estimated by omitting the router's flow, can accurately calculate the feedback rate, and the simulation experiments of various data flow environment on NS2 simulation platform, the results show that the The utilization rate of algorithm and RCP had higher network, and has a lower queue length and shorter average flow time. So this algorithm can accurately in high speed feedback rate and has shorter average flow time, application performance can improve the network. (2) analyzed the status quo and limitation TCP congestion control in wireless networks, the wireless environment has high error rate, AMID congestion control scheme, the throughput will drop sharply. The main reason is that traditional TCP congestion control mechanism is not able to distinguish the cause of packet loss, which is caused by the wireless channel or packet loss caused by network congestion, so in the actual network and no congestion occurs, but due to wireless channel errors caused by no code in the case of packet loss, TCP still reduces its sending rate. Based on the existing end-to-end bandwidth estimation of congestion control. The case has left TCP congestion control protocol shortcomings, such as performance in wireless link under low packet throughput jitter generation is too large, the bandwidth utilization rate is not high. This paper presents a bandwidth estimation feedback rate based congestion control scheme, and design a method to estimate the accurate rate of bandwidth, RCP protocol feedback congestion control algorithm is applied to the wireless network based on NS2, according to the analysis of simulation experiment, we can obtain higher throughput and stability. (3) of multi path transmission control protocol was analyzed, found that many routes and reliable transmission is very important, at the same time, the MPTCP congestion control mechanisms are discussed, with the rate feedback instead of windows based on MPTCP protocol, we propose a MPTCP congestion control scheme based on the accurate rate feedback congestion control protocol applied to each MPTCP flow. In stopper control, we can improve the throughput of each sub stream, and theoretically, the scheme can quickly utilize the remaining bandwidth and have faster convergence speed and higher throughput.

【學(xué)位授予單位】：寧波大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：TP393.06

【參考文獻(xiàn)】

相關(guān)期刊論文 前4條

1 劉正偉;文中領(lǐng);張海濤;;云計算和云數(shù)據(jù)管理技術(shù)[J];計算機研究與發(fā)展;2012年S1期

2 羅萬明,林闖,閻保平;TCP/IP擁塞控制研究[J];計算機學(xué)報;2001年01期

3 魏佳杰;郭曉金;;TCP擁塞控制技術(shù)研究[J];現(xiàn)代電子技術(shù);2009年15期

4 ;EVCP:a convergence time improved high-speed transport congestion control protocol[J];Journal of Zhejiang University(Science A:An International Applied Physics & Engineering Journal);2007年06期

，

本文編號：1720824