中南大学学报(自然科学版)

MANET中TCP拥塞控制方法综述

刘洪全1, 2,谷源涛1

(1. 清华大学 电子工程系,北京,100084;

2. 95989部队,北京,100076)

摘 要:

NET(mobile Ad-hoc NETwork) 中的关键问题。MANET具有无线信号干扰大、信道衰落严重、多跳路由、链路不对称和网络拓扑动态扩展等特性,这些特性会使传统TCP(transmission control protocol) 在发送数据过程中经历丢包率高、应答包在中间节点累计、乱序包多和往返时间抖动大等问题,容易触发TCP作出错误的拥塞控制,进行不必要的超时重传、丢包重传和减小发送窗口大小,最终严重降低端到端的吞吐量。本文调研了近年来针对MANET提出的各种拥塞控制改进方法,按照拥塞产生的原因进行分类总结,详细介绍各个算法的工作原理并比较这些方法的技术特点。发现通过跨层设计获取网络实时状态信息,进而用来辅助拥塞控制,是改进方法的主流实现方式。

关键词:

移动自组织网络拥塞控制无线多跳网络TCP

中图分类号:TN92            文献标志码:A         文章编号:1672-7207(2013)01-0156-10

Survey on TCP congestion control for MANET

LIU Hongquan1, 2, GU Yuantao1

(1. Department of Electronic Engineering, Tsinghua University, Beijing 100084, China;

2. 95989 Troops, Beijing 100076, China)

Abstract: Congestion control is a key issue in MANET. MANET has some special properties, such as large interference of wireless signal, heavy fading channel, multiple hops, unsymmetry links and node mobility, which may cause high packet loss rate, acknowledgement packets cumulating in intermediate nodes, lots of out-of-order packets and large jitter of round trip time. Consequently, improper congestion controls (e.g., unnecessary retransmission and mistakenly reducing the size of send window) are made and the throughput is very poor in MANET with traditional TCP congestion control mechanism. Many typical improved approaches proposed to overcome these difficulties in MANET were verified in this survey. In addition, the features and performances of these approaches were compared and concluded. According to this survey, real-time state information of the whole network obtained by cross-layer design is helpful to congestion control in MANET and this is the mainstream mechanism adopted by these proposed approaches.

Key words: mobile ad hoc network; congestion control; multi-hop wireless network; TCP

MANET(如图1所示)是由一组带有无线收发装置的移动终端组成的临时性自治系统[1-2]。MANET没有固定的基础设施,每个终端不仅能够移动,而且兼有路由器和主机2种功能[3]。自组织和移动的特点使得MANET易于在无基础设施的环境中迅速构建无线网络,因而在灾后重建、战场通信、车辆移动通信和科学探险等环境中得到广泛应用[2]。同时,无中心、动态拓扑、资源受限等特点[1]使得MANET必须面临各种各样的复杂通信环境。MANET的通信特点使得传统的拥塞控制方法已不再适用,这是因为传统的拥塞控制方法是针对有线网络提出,它认为网络中的丢包是由拥塞引起,并且数据包很少发生乱序现象;而MANET存在严重的信道衰落、路由中断、信道接入不公平等问题,这使得MANET中存在大量的非拥塞丢包和乱序包,并且延时抖动很大,进而容易触发传统的拥塞控制方法进行错误的拥塞控制,造成网络吞吐量下降,传输延时增加[4]。拥塞控制已成为提高MANET传输性能的关键问题[5]

图1  移动自组织网络示意图

Fig.1  Example of mobile ad-hoc network

1  MANET中的拥塞控制问题

在一个共享资源的网络中,终端竞争使用网络的带宽资源。但资源的受限性导致路由缓存已满时后续到达的数据包被直接丢掉,从而引发重传,导致更多的数据包进入网络,直到网络瘫痪。因此,必须进行拥塞控制,这也是标准TCP拥塞控制方法产生的原因。如果算法只在收发端进行速率控制,不需要其他节点的参与,则该类拥塞控制算法被称为SA(source algorithm)[6];如果算法需要网络中间设备反馈相应拥塞信息,则该类拥塞控制算法被称为LA(link algorithm)[6]。传统TCP协议针对有线网络设计,MANET的节点移动、多跳和共享无线信道等特点使得传统方法出现了很多新问题。

1.1  节点移动产生的问题

节点移动导致MANET拓扑动态变化,进而传输路由会频繁中断和恢复,合理的解决方法[5]是在路由恢复之前停止发送数据,避免浪费网络资源;在路由恢复之后立即重传丢失的数据包并且不减小拥塞控制窗口以提高网络吞吐量,维持带宽利用率。传统TCP不能分辨丢包的原因是拥塞还是路由中断,一并归结为拥塞,执行拥塞避免算法(慢启动门限设置为当前窗口一半,拥塞控制窗口设置为1),因此,在路由恢复后,发送速率从极小值逐渐增大,浪费了带宽,导致吞吐率大幅下降。

节点移动还会带来包乱序。文献[7-8]详细分析了无线链路上产生包乱序的原因,定义由链路层数据重传产生的包乱序为第Ⅰ类;由节点移动、路由切换产生的包乱序为第Ⅱ类。MANET路由中断主要引起第Ⅱ类包乱序。而在标准TCP的包乱序处理流程(如图2所示)中,假如数据包(201~400)因为路由变换到达较晚,收端在先收到(401~600),(601~800)和(801~1 000)会分别回复DupACK(duplicate acknowledgement),导致发端收到3个DupACK后重发数据包(201~400),显然这种处理方法是不恰当的且浪费资源。

图2  标准TCP处理乱序包的示意图

Fig.2  Example of conventional TCP dealing with out-of-order packets

节点移动是MANET的典型特点之一,是TCP拥塞控制方法需要重点解决的问题。

1.2  无线多跳产生的问题

无线多跳是MANET的另一典型特点,增大通信距离和易接入性的同时也带来随机丢包和信道接入公平性问题[1, 5],提高了MANET中拥塞控制改进方法的设计难度。

1.2.1  随机丢包问题

无线链路存在多径衰落和(节点移动产生的)阴影效应[1, 3-5],会产生无线丢包;同时作为无线多跳网络,MANET还存在隐藏终端。如图3所示,当A向B发送数据时,C是隐藏终端,C若向B发送数据,它会对接收节点产生电磁干扰,造成接收失败,产生碰撞丢包;另外节点移动引起不可恢复路由中断时也会产生丢包。上述众多原因造成MANET中的丢包概率远大于有线网络乃至其他的一些无线网络。标准TCP误把随机丢包当成网络拥塞从而进入慢启动,导致端到端传输性能下降。

图3  隐藏终端示意图

Fig.3  Example of hidden terminal

1.2.2  信道接入公平性问题

多个终端共享网络中的无线链路,所以公平合理地在各个终端间分配网络资源是一个重要的问题。由于MANET中存在隐藏终端和暴露终端(如图4所示)等问题,现有IEEE 802.11协议的接入控制并不公平,而多跳的特点使得MANET中信道公平分配问题不仅需要考虑与节点相邻的一跳节点,还需要考虑2跳甚至3跳的节点,这加大了拥塞控制的难度。另外MANET中存在着多种不对称因素[1, 5],如往返路径、带宽、丢包率、路由等,导致ACK包被积压在中间瓶颈节点,不能及时到达发端,触发传统拥塞控制方法突发数据,增加拥塞概率。

图4  暴露终端示意图

Fig.4  Example of exposed terminal

MANET网络环境复杂多变,节点移动引起拓扑动态变化,链路的传输路径、传输延时也随之频繁变化,使用传统TCP的拥塞控制会产生很多问题,严重降低网络性能。因此如何整合MAC和路由等各种可能获得的信息,设计给出适应于MANET网络的拥塞控制算法变得至关重要。

2  MANET中的拥塞控制改进算法

2.1  针对节点移动产生的拥塞问题的改进方法

因为节点的移动性,MANET中经常发生路由中断和恢复[5],这给传统TCP带来一系列问题,如路由中断导致数据包重发、RTO(Retransmission Timeout)增长太快等;路由恢复后新路径的特性可能发生变化,原有的拥塞窗口大小、慢启动阈值等参数已不再适合。路由中断产生的问题是拥塞控制的关键所在,因而这方面出现了大量的改进方法。

TCP-F(TCP-Feedback)[9]和TCP-ELFN(TCP-explicit link failure notification)[10]是较早出现,用以解决MANET路由中断引起拥塞控制问题的方法。TCP-F方法由Chandran等提出,其主要思想是当路由中断引起非拥塞丢包或RTO发生后,停止TCP自身的拥塞控制机制。路由中断和恢复事件由RFN(route failure notification,路由失败通知)和RRN(route re- establishment notification,路由恢复通知)获得,这2个通知由中间节点产生并转发。当发端收到RFN后进入“snooze”状态,锁定计时器、窗口大小等参数,当收到RRN后再行恢复,这样就避免TCP在路由中断和恢复的过程中采取不恰当的操作。Holland等[10]分析了ELFN(explicit link failure notification) 技术应用在TCP上的性能,提出TCP-ELFN算法,通过低层反馈感知链路中断或者路由中断,发端进入类似TCP-F中“snooze”状态的“standby”模式,但是相比TCP-F算法,TCP-ELFN没有引入额外的控制包,而是通过路由协议的通知包或者ICMP(internet control message protocol)目的不可达包来得知链路中断,通过收到发送队列中第一个包的确认包得知路由恢复。TCP-F和TCP-ELFN算法具有里程碑意义,为后继算法广泛借鉴。

延续上述的反馈思想,有学者提出了TCP-BuS (TCP with buffering capability and sequence information)[11],ENIC(enhanced inter-layer communication and control)[12],TCP-RC(TCP- ReComputation)[13],ATCP(TCP for mobile ad hoc networks)[14]等算法。Kim等拓展路由直接通知的思想提出了TCP-BuS算法,该算法由路由协议通告上层发生路由中断,中间节点发送RRN、发端发送探测包得知路由恢复;采用选择式应答机制;在中间节点缓存一些数据包,路由恢复后能快速重传丢包,避免这些包发生重传超时。ENIC由Sun和Man提出,采用类似ELFN的路由中断处理策略,并综合了TCP DACK(delay ACK)[15] SACK (selective ACK)[16]方法。在ENIC中发送端和接收端都会收到路由中断通知并进入锁定状态,在路由恢复后根据新建路由计算新的重传超时值,ENIC是第一个考虑重建路由新特性的算法。TCP-RC在ELFN的基础上利用新建路由长度和RTT(round-trip time)重新计算TCP的拥塞控制窗口和慢启动门限,在路由恢复时能够最大限度地利用带宽,减少浪费。ATCP是为了解决MANET中路由中断、区分拥塞和其他种类丢包、长时间不连接等问题提出的,它采用类似TCP-ELFN使用的探测包方法,探测链路当前状态,当拥塞发生时中间节点产生ECN(explicit congestion notification) 包,若发端未收到ECN则认为非拥塞引起丢包,锁定当前参数,当链路恢复时再行恢复。

不同于上述利用低层或者其他节点的反馈解决路由中断问题的思路,Fixed RTO[17],TCP-DOOR(TCP with detection of out-of-order and response)[18],TCP-R (TCP for reordering)[19]和ABRA(adaptive back-off response approach)[20]是端到端的方法。端到端方法,如图5所示。直接在发端或收端的TCP协议或传输层上做改进,保持了TCP连接的端到端语义,不需要修改中间设备,易于在异质中实施。Dyer和Boppana通过比较3种MANET路由协议下不同TCP改进方法的性能,发现路由中断引起RTO呈幂次增加,进而提出了Fixed RTO方法,主要方法是在发生初次重传后保持固定的RTO,仿真表明该方法可以获得与TCP-ELFN方法相近的性能。TCP-DOOR通过观察乱序包到达顺序来识别路由中断和路由变化。发端有2种途径得知发生乱序:一是收端发现包乱序到达后通知发端;二是发端通过监测ACK包到达顺序得知是否发生乱序。当检测到发生乱序后发端采取2步措施,首先发端临时中断TCP的拥塞控制算法并使TCP的各个参数保持不变甚至回退到之前的某个状态;随后当路由恢复时,TCP立刻恢复到中断前的状态,就像没有发生过路由中断一样。TCP-R改进方法在发端收到3个DupACK后延迟一段时间进入拥塞避免和快速重传,而在这段时间内有可能被路由变化延误的包已经到达收端并且发端也收到了该包的ACK(如图1中的ACK(1001)包),这样就可以避免发端进行错误的拥塞控制。ABRA也采用锁定TCP参数的方法,但是它不需要下层反馈信息:发生RTO时锁定参数,在重发数据包收到回复后解除锁定,ABRA可使RTO不呈指数性增长。Fard等[21]提出算法来合理调整RTO进而提高TCP性能,该算法通过队列缓存的变化情况来区分拥塞丢包和链路中断丢包,发生丢包时队列使用呈增长趋势表明是拥塞丢包,队列长度基本不变表明是链路中断丢包,然后,算法对比新旧两条路径的传输延时和跳数决定是增加还是减少RTO,最终使TCP合理利用带宽资源,增加吞吐量。另外,路由中断引起的ACK包丢失会引发不必要的重传,Kim等[22]把ACK序列号封装到路由协议数据包中,在收端寻找新路由时发给它,从而解决了因路由中断ACK包丢失引发的重传问题。

图5  端到端改进方法

Fig.5  End-to-end improved mechanism

IBOM等[23]对TCP-F,TCP-ELFN,Fixed RTO,ATCP和TCP-DOOR进行性能分析时发现,在多TCP连接时前3种改进方法吞吐性能相近,优于ATCP,但远差于TCP-DOOR。

前述改进方法都是在路由中断发生后进行处理,而PR(preemptive routing)[24],SSBLM(signal strength based link management)[25]和AF(atra framework)[26]等改进方法则试图在路由中断发生之前进行处理,目的是避免发生中断或减少中断持续时间。前两者通过监测路由线路上节点的接收信号强度进行预测,若低于某一阈值则认为路由即将中断进而提前寻找建立新路由。Anantharaman等[26]指出ELFN方法只能解决路由中断引起的部分问题,进而提出了基于DSR(dynamic source routing)协议的AF方法,目标是尽量减小路由中断带来的影响,并提出Symmetric Route Pinning,Route Failure Prediction和Proactive Route Errors 3种算法进行预测,同时在中断发生前将消息快速通知给路由上的其他节点。

2.2  针对无线多跳产生的拥塞问题的改进方法

MANET多跳和共享信道带来随机丢包和传输公平性问题,针对这2个问题也提出了很多改进方法。

2.2.1  随机丢包问题

无线多跳网络中因为碰撞、链路中断等原因经常会发生随机丢包,若像传统TCP那样把随机丢包当作拥塞丢包处理,则将严重降低吞吐能力。在有线无线混合网络中已经做了大量的工作来区分随机丢包和拥塞丢包,本节主要介绍针对MANET随机丢包问题的改进方法。

TCP/RCWE(TCP with restricted congestion window Enlargement)[27]和ADTCP(Ad hoc TCP)[28]是较早出现的解决MANET中随机丢包问题的方法。TCP/RCWE在ELFN基础上增加RCWE算法处理随机丢包,该算法实时观测RTO,当RTO增大时不增加拥塞窗口,当RTO减小或不变时按照TCP自身规则增加拥塞窗口,仿真表明该算法的拥塞控制窗口较小,比TCP-ELFN吞吐性能更好,丢包率更低。Fu等[28]认为MANET中传统TCP性能不佳的原因是干扰影响了某些反映网络状态的参数的准确性,从而提出了ADTCP方法改进探测误比特丢包的可靠性。该方法在收端通过2个参数判断拥塞发生情况:一是2个相邻数据包的到达时间间隔,间隔增大意味着发生拥塞;另一个是过去短时间(如10 s)内的吞吐量,拥塞发生时吞吐量必减小;同时,ADTCP监测包到达的顺序来判断是否发生路由中断和变化,收端把计算得到的参数实时发送给发端。发端和收端根据这些信息做一些更合适MANET的拥塞控制,从而提高吞吐量。Gajjar等[29]对ADTCP的CWL(congestion window limit)进行修改,提出I-ADTCP(improved-ADTCP)方法,把ADTCP的性能提高了10%~30%。de Oliveira等[30]通过RTT来区分拥塞丢包和随机丢包,提出了Edge-based的方法,当探测到随机丢包后,不触发TCP拥塞控制算法;当发生超时或长时间没有收到包时认为路由中断,进入类似ELFN的探测模式定期发送包以探测路由是否 恢复。

Fu等[31]发现在MANET中较小的TCP拥塞窗口能够减少包碰撞的可能性,进而减少随机丢包,增加TCP吞吐量。基于这个思想出现了很多方法,典型的有SCA(slow congestion avoidance)[32]、FeW(fractional window increment)[33]、RBCC(rate-based congestion control)[34]和C3TCP(cross-layer congestion control TCP)[35]等。SCA与FeW方法相似,都是控制TCP窗口的增加速度使其慢于传统TCP。RBCC在控制TCP窗口增加速率的基础上增加了一个漏桶算法:在包头增加一个反馈域,每个中间节点计算自己允许的最大传输速率,若小于转发包中反馈域的值则更新它,这样发端收到该信息后即可精确控制发送速度,避免拥塞。C3TCP与RBCC类似,也在包头增加一个反馈域,用来记录中间节点的最小剩余带宽和数据包的累计延时,收端获得这些信息后调整接收窗口大小,随后发端通过ACK也会得到上述信息,进而动态控制窗口大小,提高吞吐量。

Lien等[36]提出TCP Muzha方法解决随机丢包问题,提出MRAI(minimum data rate adjustment index)指标,用来表示节点给某TCP流提供的最大可用带宽,每个中间节点计算自己的MRAI值,与发端添加在TCP头部自定义区域的值进行比较,如果小于则替代之,收端收到后再通过ACK反馈给发端,发端即可根据此信息进行流控,减少数据冲突。类似后面提到的Split TCP[37]方法,Lien等[38]又提出一种Hop-by-Hop TCP的方法,本跳节点在传输包头中增加自己的IP信息,下跳节点收到包后产生单独的确认消息给本跳节点,这样便在跳与跳之间建立可靠传输,丢包发生时能够以最快速度重传。

MANET网络环境复杂多变,仅仅根据传输层获得的信息改进拥塞控制方法的效果很不理想,为了准确获得更多反映网络状况的信息,有必要进行跨层设计,Lochert[5]指出跨层设计是拥塞控制的一个主要改进方向。MANET中TCP性能差的本质原因是在传输层不能准确判断网络状况导致采取不恰当的拥塞控制,如图6所示,通过跨层设计可以在上下协议层间共享信息,可从链路层获得链路丢包率信息和信号强度信息等,可从网络层获得路由缓存信息和路径信息等,这些信息都有助于拥塞控制算法准备判断网络状况。Giovanidis等[39-40]以NUM(network utility maximization) 为目标,统筹考虑传输层、网络层和MAC层的优化算法,通过合理分配发射功率和发送速率,配合相应的路由算法,减少传输碰撞,降低信道衰落的影响,最终减少随机丢包,达到NUM。Yu等[41]联合传输层和MAC层提出一种算法,控制链路上每个节点的随机接入概率,从而合理调度链路上不同流的发送速率,减少随机丢包,减轻拥塞,最终使所有源节点发送速率之和达到最大。

图6  跨层共享信息

Fig.6  Share information by cross layer

2.2.2  信道接入公平性问题

无线网络中若干节点共享同一链路,给MAC层接入控制带来很大困难,很容易产生TCP流间不公平、流内传输冲突等问题。

COPAS(contention-based path selection)[42]和NRED(neighborhood random early dropping)[43]是解决MANET中不公平问题的典型方法。COPAS在建立路由时选取2个不同的路径,分别用来传输上行和下行数据,从而解决双向数据接入信道冲突问题。NRED通过侦听邻居节点的收发信号来分析信道的利用率,进而在本节点生成一个虚拟的数据队列,该队列反映了所有邻居节点传输数据量之和,当队列长度超过某一阈值时产生一个丢包概率并通知给邻居节点,邻居节点根据通知计算自己的丢包概率,并以该概率丢弃新接收的数据包,形成类似RED(Random Early Dropping) 的丢包处理。LRED[44]通过包重传次数感知的信道冲突情况,然后微调无线链路的丢包率使TCP能够维持一个合理的窗口大小,减小发生包碰撞的可能性,增加吞吐量。Zhai等[45-46]提出OPET(Optimum Packet scheduling for Each Traffic Flow)改进方法,采用4项技术解决流间和流内的冲突问题,第1项是增加接收数据节点的接入优先级从而优先传输下行数据,解决流内双向传输间的矛盾;第2项类似802.11的RTS/CTS算法,只有在下一跳给出允许发送通知后才开始发送上行数据;第3项限制发送速率,防止同一TCP流占用过多的带宽;最后一项称作基于流的循环通知,用来防止某一跳数据流过度占用信道。之前已有学者[47-48]在MANET中使用DACK[15]算法,D2ACK (Dynamic Delayed ACK)[49]方法扩展了DACK思想,只有在收到d个数据单元或收端自定义计时器超时后才发送ACK,以减少ACK数量,减少双向数据包之间的冲突,同时也能缓解包乱序的影响,d的取值根据网络情况动态调整,一般为2~4。在ASC(alleviating self-contention)[50]方法中,Berger等基于802.11的RTS/CTS(request to send/clear to send)提出了QE(quick exchange)和FF(fast forward) 2种算法以减少双向传输的冲突,提高发送速率。QE在一次交换RTS/CTS信息时发送2个数据包;FF把ACK附加在RTS包上加快下行数据的发送速率。

Kopparty等[37]以一种完全不同的思路提出了Split TCP方法,把拥塞控制和端到端的可靠传输2种功能分开。中间节点自动成为TCP代理,把数据发送给下一个代理或者收端,2个代理间采用独立的发送应答机制保证传输的可靠性。因为可靠传输跳数变少,所以在一定程度上缓解了双向数据接入信道冲突问题。

MANET中发端和收端间存在多条连通路径,可以通过优化路由协议解决流间的不公平问题,减小发生拥塞的几率。EDAODV(early detection congestion and control routing protocol)[51]路由算法通过实时监测路由节点缓存大小来预测是否将发生拥塞,并在拥塞发生前通知前后节点,促使它们寻找并建立新路径;TASR(timestamp based multipath source routing)[52]是一种基于DSR的多径路由算法,它把数据流均衡分配在多条路径上,防止某些数据流过度占用单一路径带宽,减小发生拥塞的几率。

从应用角度看,MANET很可能主要承载视频和语音业务[53-56],因此在传输层使用UDP协议。在802.11e标准中,基于UDP的多媒体传输具有高优先级,使得TCP数据不能及时传输,进而使得TCP产生很多超时。Mbarushimana等[54-55]中提出E-TCP方法,通过增加TCP中ACK的优先级以及根据信道冲突情况调整重传计时器来解决这个问题。鉴于UDP数据流没有拥塞控制,Premalatha等[57]提出1种UDP拥塞控制方法,利用收端的反馈信息控制UDP的发送速率,使TCP流和UDP流公平竞争,提高TCP的吞吐量。Calafate等[58-59]针对MANET中的视频传输问题,提出DACME(distributed admission control for MANET’s)算法,该算法运行在视频服务器上,周期性地向目的节点发送UDP探测包,目的节点把需要的带宽和延时等信息通过应答返回,服务器统计各数据流的传输信息并当某数据包不能满足使用需求时拒绝发送该包;对通过的数据包赋予不同的优先级,统筹规划带宽分配,提高公平性;该算法采用跨层设计。

2.3  拥塞控制新方法

传统TCP的拥塞控制方式不适合MANET,因而涌现出大量的修正方法。同时,也有学者摈弃了修修补补的思路,提出几种针对MANET特点的新型拥塞控制协议。纵然难以与现有网络兼容,但是在某些特殊场景中(如野外抢险需要建立一个封闭网络),这些新协议具有重要的现实意义和实用价值。

最早提出的新协议是EXACT(explicit rate-based flow control)[60-61],基于速率进行拥塞控制。发端在数据包头部存放2个速率值,其中ER(explicit rate)表示允许的发送速率,CR(current rate)表示当前的发送速率。中间路由节点记录某数据流的CR值,根据自己的带宽,以公平分配为原则计算本节点可分配给该流的速率,若小于包头中的ER值则更新它,收端将收到的ER反馈给发端,发端即可根据该值控制发送速率。EXACT提出安全窗口的概念,发端不允许未应答的数据包数大于安全窗口,防止发端在路由中断时过度发送数据。EXACT没有重传计时器,而是采用SACK方式,只有当未应答序号与最大应答序号相差较大时才重传该序号的包。

另一个与EXACT类似的协议是ATP(ad-hoc transport protocol)[62],它同样没有重传计时器,拥塞控制和可靠传输分开工作。启动时ATP发送探测包估计网络带宽,计算出初始发送速率和延时,并以此速率发送数据,同时延时值被封装在包头中。中间节点计算数据包的平均延时,若大于包头记录的延时值则更新后者,这样收端即得到传输路径上的最大延时并将此信息周期性地反馈给发端,发端基于该信息进行流速控制。

与前面基于速率进行拥塞控制不同,TPA (transport protocol for ad hoc networks)[63]沿用了TCP基于窗口控制发送速率的机制。该方法的出发点是尽可能的减少ACK的数量。发端把K个数据包编成一个块,收端周期性地反馈描述该块数据接收情况的ACK,而不是对每个接收包发送一个ACK,这样相对TCP而言减少了ACK的数量。发端在一块数据都成功接收后才会启动下一个数据块的发送。TPA使用ELFN算法处理路由中断,通过窗口方式进行拥塞控制,正常时发送窗口设定为最大值,一般为2或3,收到ELFN通知后窗口减小为1。

XRCC(explicit rate-based congestion control)[56]是针对流媒体传输问题提出的基于速率的拥塞控制新方法,中间路由节点把检测到的丢包信息和速率信息附在数据包上,收端在收到这些信息后再通过应答包反馈给发端,发端根据这些信息就能区分拥塞丢包和无线丢包,进而合理控制发送速率,提高TCP吞吐量。

3  总结与展望

为提高拥塞控制的准确性,使协议栈灵活地适应无线移动环境,利用中间节点反馈网络状态信息,通过跨层设计在各层之间共享特定信息,是MANET中拥塞控制改进方法的主流研究方向。跨层设计打破了传统的层次化结构,在传输层可获得链路层和网络层的信息,同时可充分利用它们提供的服务进行拥塞控制,如在RTS/CTS握手时传输下行数据,根据路由反馈的链路信息提前进行流量控制等。表1所示比较结果也验证了这一结论。表1比较了前述各种改进方法的主要技术特点(如表1所示,其中“√”表示使用或支持某项技术,“×”表示不使用或不支持,“-”表示不能明确看出是否使用或支持)。从表1可看出:多数改进方法都是LA型,多利用了中间节点反馈信息以及采用跨层设计提高MANET的网络性能。

表1  各改进方法的共性

Table 1  General characters of proposed mechanisms

MANET目前已经被扩展应用到各种不同的网络架构当中,例如vehicular ad hoc networks(VANETs),wireless mesh networks(WMNs),和wireless sensor networks(WSNs)。这些网络的新特点(如高速移动、能量受限等)以及应用程序的新需求对拥塞控制提出了新的要求。未来拥塞控制改进算法的研究需要能够针对这些具体需求提出解决方案。

参考文献:

[1] Narsimha1 V B, Sujatha B, SampathKumar T. A survey of wireless mobile ad-hoc networks (MANET)[J]. International Journal of Science and Advanced Technology, 2011, 1(5): 189-192.

[2] Mamatha G S, Sharma S C. Network layer attacks and defense mechanisms in MANETS: A survey[J]. International Journal of Computer Applications, 2010, 9(9): 12-17.

[3] Agarwal R, Motwani M. Survey of clustering algorithms for MANET[J]. International Journal on Computer Science and Engineering, 2009, 1(2): 98-104.

[4] Jawandhiya P M. A survey of mobile ad hoc network attacks[J]. International Journal of Engineering Science and Technology, 2010, 2(9): 4063-4071.

[5] Lochert C, Scheuermann B, Mauve M. A survey on congestion control for mobile ad-hoc networks[J]. Wiley Wireless Communications and Mobile Computing, 2007, 7(5): 655-676.

[6] Liu G, Du J. Performance analysis of link algorithm in congestion control[C]//Advanced Computer Theory and Engineering (ICACTE). Chengdu: IEEE Computer Society, 2010: V3-84-V3-87.

[7] Lai C, Leung K C, Li V O K. TCP-NCL: A unified solution for TCP packet reordering and random loss[C]//IEEE 20th International Symposium on Personal, Indoor and Mobile Radio Communications. Tokyo: IEEE Computer Society, 2009: 1093-1097.

[8] Lai C, Leung K C, Li V O K. Enhancing wireless TCP: A serialized-timer approach[C]//Proceedings of IEEE INFOCOM. San Diego: IEEE Computer Society, 2010: 1-5.

[9] Chandran K, Raghunathan S, Venkatesan S, et al. A feedback based scheme for improving TCP performance in ad-hoc wireless networks[C]//Proceedings of the 18th International Conference on Distributed Computing Systems. Amsterdam: IEEE Computer Society, 1998: 472-479.

[10] Holland G, Vaidya N H. Analysis of TCP performance over mobile ad hoc networks[C]//Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking. Seattle: IEEE Computer Society, 1999: 219-230.

[11] Kim D, Toh C K, Choi Y. TCP-Bus: Improving TCP performance in wireless ad-hoc networks[C]//Proceedings of the IEEE International Conference on Communications. New Orleans, 2000: 1707-1713.

[12] Sun D, Man H. ENIC—An improved reliable transport scheme for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. San Antonio, TX: IEEE Computer Society, 2001: 2852-2856.

[13] Zhou J, Shi B, Zou L. Improve TCP performance in ad hoc network by TCP-RC[C]//Proceedings of the 14th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications. Beijing: IEEE Computer Society, 2003: 216-220.

[14] Liu J, Singh S. ATCP: TCP for mobile ad hoc networks[J]. IEEE Journal on Selceted Areas in Communications (J-SAC), 2001, 19(7): 1300-1315.

[15] RFC 1122, Requirements for Internet Hosts- Communication Layers[S].

[16] RFC 2018, TCP Selective Acknowledgment Options[S].

[17] Dyer T D, Boppana R V. A comparison of TCP performance over three routing protocols for mobile ad hoc networks[C]// Proceedings of the 2nd ACM International Symposium on Mobile Ad Hoc Networking & Computing. Long Beach: ACM New York, 2001: 56-66.

[18] Wang F, Zhang Y. Improving TCP performance over mobile ad-hoc networks with out-of-order detection and response[C]// Proceedings of the 3rd ACM International Symposium on Mobile Ad Hoc Networking &Computing. Lausanne: ACM New York, 2002: 217-225.

[19] Sun W, Wen T, Guo Q. A novel protocol for mobile-induced packet reordering in mobile ad hoc networks[C]//Information Science and Engineering. Shanghai: IEEE Computer Society, 2008: 626-631.

[20] Ghanem T F, Elkilani W S, Hadhoud M M. Improving TCP performance over mobile ad hoc networks using an adaptive backoff response approach[C]//Proceedings of International Conference on Networking and Media Convergence. Cairo: IEEE Computer Society, 2009: 16-21.

[21] Fard M A K, Bakar K A, Karamizadeh S, et al. Improve TCP performance over mobile ad hoc network by retransmission timeout adjustment[C]//Communication Software and Networks (ICCSN), 2011 IEEE 3rd International Conference on. Xi’an, China: IEEE Computer Society, 2011: 437-441.

[22] Kim D, Toh C K, Yoo H. The impact of spurious retransmissions on TCP performance in aD hoc mobile wireless networks[C]// Personal, Indoor and Mobile Radio Communications. Athens: IEEE Computer Society, 2007: 1-5.

[23] Ibom U O. TCP performance over MANET[C]//Information Networking. Busan: IEEE Computer Society, 2008: 1-5.

[24] Goff T, Abu-ghazaleh N B, Phatak D S, et al. Preemptive routing in ad hoc networks[J]. ElsevierJournal of Parallel and Distributed Computing, 2003, 63(2): 123-140.

[25] Klemm F, Ye Z, Krishnamurthy S V, et al. Improving TCP performance in ad hoc networks using signal strength based link management[J]. Elsevier Ad Hoc Networks, 2005, 3(2): 175-191.

[26] Anantharaman V, Park S J, Sundaresan K, et al. TCP performance over mobile ad-hoc networks: Aquantitative study[J]. Wireless Communications and Mobile Computing, 2004, 4(2): 203-222.

[27] Günes M, Vlahovic D. The performance of the TCP/RCWE enhancement for ad-hoc networks[C]//Proceedings of the 7th IEEE International Symposium on Computers and Communication. Taormina, Giardini Naxos: IEEE Computer Society, 2002: 43-48.

[28] Fu Z, Greenstein B, Meng X, et al. Design and implementation of a TCP-friendly transport protocol for ad hoc wireless networks[C]//Proceedings of the 10th IEEE International Conference on Network Protocols. Paris: IEEE Computer Society, 2002: 216-225.

[29] Gajjar S, Gupta H M. Improving performance of ad hoc TCP in mobile ad hoc networks[C]//India Conference. Kanpur: IEEE Computer Society, 2008: 144-147.

[30] de Oliveira R, Braun T, Heissenb¨uttel M. An edge-based approach for improving TCP in wireless mobile ad hoc networks[C]//Proceedings of the Conference on Design, Analysis and Simulation of Distributed Systems. Orlando: The Society for Modeling and Simulation International (SCS), 2003: 172-177.

[31] Fu Z, Zerfos P, Luo H, et al. The impact of multihop wireless channel on TCP throughput and loss[C]//Proceedings of the 22nd Annual Joint Conference of the IEEE Computer and Communications Societies. San Francisco California: IEEE Computer Society, 2003: 1744-1753.

[32] Papanastasiou S, Ould-khaoua M. TCP congestion window evolution and spatial reuse in MANETs[J]. Journal of Wireless Communications and Mobile Computing, 2004, 4(6): 669-682.

[33] Nahm K, Helmy A, Kuo C C J. TCP over multihop 802.11 networks: Issues and performance enhancement[C]//Proceedings of the 6th ACM International Symposium on Mobile Ad Hoc Networking and Computing, Urbana-Champaign: ACM New York, 2005: 277-287.

[34] Zhai H, Chen X, Fang Y. Rate-based transport control for mobile ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. New Orleans: IEEE Computer Society, 2005: 2264-2269.

[35] Kliazovich D, Granelli F. Cross-layer congestion control in ad hoc wireless networks[J]. Ad Hoc Networks, 2006, 4(6): 687-708.

[36] Lien Y N, Hsiao H C. A new TCP congestion control mechanism over wireless ad hoc networks by router-assisted approach[C]// Distributed Computing Systems Workshops. Toronto: IEEE Computer Society, 2007: 84-84.

[37] Kopparty S, Krishnamurthy S, Faloutsos M, et al. Split-TCP for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. Taipei: IEEE Computer Society, 2002: 138-142.

[38] Lien Y N, Hsiao H C. Hop-by-Hop TCP over MANET[C]// Asia-Pacific Services Computing Conference. Yilan: IEEE Computer Society, 2008: 1150-1155.

[39] Giovanidis A, Stanczak S. Retransmission aware congestion control and distributed power allocation in MANETs[C]// Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks. Seoul: IEEE Computer Society, 2009: 1-10.

[40] Papandriopoulos J, Dey S, Evans J. Optimal and distributed protocols for cross-layer design of physical and transport layers in MANETs[J]. IEEE/ACM Transactions on Networking, 2008, 16(6): 1392-1405.

[41] Yu Y Q, Giannakis G B. Cross-layer congestion and contention control for wireless ad hoc networks[J]. IEEE Transactions on Wireless Communications, 2008, 7(1): 37-42.

[42] Cordeiro C D A, Das S R, Agrawal D P. COPAS: Dynamic contention-balancing to enhance the performance of TCP over multi-hop wireless networks[C]//Proceedings of the 10th International Conference on Computer Communication and Networks (IC3N). Miami: IEEE Computer Society, 2002: 382-387.

[43] Xu K, Gerla M, Qi L, et al. TCP unfairness in ad hoc wireless networks and a neighborhood RED solution[J]. Wireless Networks, 2005, 11(4): 383-399.

[44] Fu Z, Luo H, Zerfos P, et al. The impact of multihop wireless channel on TCP performance[J]. IEEE Transactions on Mobile Computing, 2005, 4(2): 209-221.

[45] Zhai H, Wang J, Fang Y. Distributed packet scheduling for multihop flows in ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. Atlanta: IEEE Computer Society, 2004: 1081-1086.

[46] Zhai H, Chen X, Fang Y. Alleviating intra-flow and inter-flow contentions for reliable service in mobile ad hoc networks[C]// Proceedings of the IEEE Military Communications Conference. Monterey: IEEE Computer Society, 2004: 1640-1646.

[47] Dyer T D, Boppana R V. A comparison of TCP performance over three routing protocols for mobile ad hoc networks[C]// Proceedings of the 2nd ACM International Symposium on Mobile Ad Hoc Networking & Computing. Long Beach: ACM New York, 2001: 56-66.

[48] Sun D, Man H. ENIC—An improved reliable transport scheme for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. San Antonio: IEEE Computer Society, 2001: 2852-2856.

[49] Altman E, T. Novel delayed ACK techniques for improving TCP performance in multihop wireless networks[C]// Proceedings of the IFIP-TC6 8th International Conference on Personal Wireless Communications. Venice: Springer, 2003: 237-250.

[50] Berger D, Ye Z, Sinha P, et al. TCP-friendly medium access control for ad-hoc wireless networks: Alleviating self- contention[C]//Proceedings of the 1st IEEE International Conference on Mobile Ad-hoc and Sensor Systems. Fort Lauderdale: IEEE Computer Society, 2004: 214-223.

[51] Kumaran T S, Sankaranarayanan V. Early detection congestion and control routing in manet[C]//Seventh International Conference On Wireless And Optical Communications Networks (WOCN). Colombo: IEEE Computer Society, 2010: 1-5.

[52] Yerajana R, Sarje A K. A timestamp based multipath source routing protocol for congestion control in MANET[C]//Advance Computing Conference. Patiala: IEEE Computer Society, 2009: 32-34.

[53] Shrivastava L, Tomar G S, Bhadauria S S. A survey on congestion adaptive routing protocols for mobile ad-hoc networks[J]. International Journal of Computer Theory and Engineering, 2011, 3(2): 189-196

[54] Mbarushimana C, Shahrabi A. E-TCP: Enhanced TCP for IEEE802.11e mobile ad hoc networks[C]//15th International Conference on Parallel and Distributed Systems (ICPADS). Shenzhen: IEEE Computer Society, 2009: 632-639.

[55] Mbarushimana C, Shahrabi A. Resource efficient TCP: Reducing contention-induced spurious timeouts in QoS-aware MANETs[C]//Advanced Information Networking and Applications. Okinawa: IEEE Computer Society, 2008: 1070-1077.

[56] Kazi C R, Syed F H. Explicit rate-based congestion control for multimedia streaming over mobile ad hoc networks[J]. International Journal of Electrical & Computer Sciences, 2010, 10(4): 28-40.

[57] Premalatha J, Balasubramanie P. Enhancing quality of service in MANETS by effective routing[C]//Wireless Communication and Sensor Computing. Chennai: IEEE Computer Society, 2010: 1-5.

[58] Calafate C T, Malumbres M P, Oliver J, et al. QoS support in MANETs: A modular architecture based on the IEEE 802.11e technology[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2009, 19(5): 678-692.

[59] Chaparro P A, Alcober J, Monteiro J, et al. Supporting scalable video transmission in MANETs through distributed admission control mechanisms[C]//18th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP). Pisa: IEEE Computer Society, 2010: 238-245.

[60] Chen K, Nahrstedt K. EXACT: An explicit rate-based flow control framework in MANET (extended version)[R]. Urbana: University of Illinois at Urbana-Champaign, 2002.

[61] Chen K, Nahrstedt K, Vaidya N. The utility of explicit rate-based flow control in mobile ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. Atlanta: IEEE Computer Society, 2004: 1921-1926.

[62] Sundaresan K, Anantharaman V, Hsieh H Y, et al. ATP: A reliable transport protocol for ad-hoc networks[C]//Proceedings of the 4th ACM International Symposium on Mobile Ad Hoc Networking &Computing. Annapolis: ACM New York, 2003: 64-75.

[63] Anastasi G, Ancillotti E, Conti M, et al. TPA: A transport protocol for ad hoc networks[C]//Proceedings of the 10th IEEE International Symposium on Computers and Communication. La Manga: IEEE Computer Society, 2005: 51-56.

(编辑  何运斌)

收稿日期:2011-11-29;修回日期:2012-02-29

基金项目:国家自然科学基金联合资助重点项目(U0835003);国家自然科学基金资助项目(60872087)

通信作者:谷源涛(1976-),男,河北保定人,副教授,从事无线网络、信源编码、图像处理研究;电话:010-62783525;E-mail: gyt@tsinghua.edu.cn

摘要:拥塞控制是MANET(mobile Ad-hoc NETwork) 中的关键问题。MANET具有无线信号干扰大、信道衰落严重、多跳路由、链路不对称和网络拓扑动态扩展等特性,这些特性会使传统TCP(transmission control protocol) 在发送数据过程中经历丢包率高、应答包在中间节点累计、乱序包多和往返时间抖动大等问题,容易触发TCP作出错误的拥塞控制,进行不必要的超时重传、丢包重传和减小发送窗口大小,最终严重降低端到端的吞吐量。本文调研了近年来针对MANET提出的各种拥塞控制改进方法,按照拥塞产生的原因进行分类总结,详细介绍各个算法的工作原理并比较这些方法的技术特点。发现通过跨层设计获取网络实时状态信息,进而用来辅助拥塞控制,是改进方法的主流实现方式。

[1] Narsimha1 V B, Sujatha B, SampathKumar T. A survey of wireless mobile ad-hoc networks (MANET)[J]. International Journal of Science and Advanced Technology, 2011, 1(5): 189-192.

[2] Mamatha G S, Sharma S C. Network layer attacks and defense mechanisms in MANETS: A survey[J]. International Journal of Computer Applications, 2010, 9(9): 12-17.

[3] Agarwal R, Motwani M. Survey of clustering algorithms for MANET[J]. International Journal on Computer Science and Engineering, 2009, 1(2): 98-104.

[4] Jawandhiya P M. A survey of mobile ad hoc network attacks[J]. International Journal of Engineering Science and Technology, 2010, 2(9): 4063-4071.

[5] Lochert C, Scheuermann B, Mauve M. A survey on congestion control for mobile ad-hoc networks[J]. Wiley Wireless Communications and Mobile Computing, 2007, 7(5): 655-676.

[6] Liu G, Du J. Performance analysis of link algorithm in congestion control[C]//Advanced Computer Theory and Engineering (ICACTE). Chengdu: IEEE Computer Society, 2010: V3-84-V3-87.

[7] Lai C, Leung K C, Li V O K. TCP-NCL: A unified solution for TCP packet reordering and random loss[C]//IEEE 20th International Symposium on Personal, Indoor and Mobile Radio Communications. Tokyo: IEEE Computer Society, 2009: 1093-1097.

[8] Lai C, Leung K C, Li V O K. Enhancing wireless TCP: A serialized-timer approach[C]//Proceedings of IEEE INFOCOM. San Diego: IEEE Computer Society, 2010: 1-5.

[9] Chandran K, Raghunathan S, Venkatesan S, et al. A feedback based scheme for improving TCP performance in ad-hoc wireless networks[C]//Proceedings of the 18th International Conference on Distributed Computing Systems. Amsterdam: IEEE Computer Society, 1998: 472-479.

[10] Holland G, Vaidya N H. Analysis of TCP performance over mobile ad hoc networks[C]//Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking. Seattle: IEEE Computer Society, 1999: 219-230.

[11] Kim D, Toh C K, Choi Y. TCP-Bus: Improving TCP performance in wireless ad-hoc networks[C]//Proceedings of the IEEE International Conference on Communications. New Orleans, 2000: 1707-1713.

[12] Sun D, Man H. ENIC—An improved reliable transport scheme for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. San Antonio, TX: IEEE Computer Society, 2001: 2852-2856.

[13] Zhou J, Shi B, Zou L. Improve TCP performance in ad hoc network by TCP-RC[C]//Proceedings of the 14th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications. Beijing: IEEE Computer Society, 2003: 216-220.

[14] Liu J, Singh S. ATCP: TCP for mobile ad hoc networks[J]. IEEE Journal on Selceted Areas in Communications (J-SAC), 2001, 19(7): 1300-1315.

[15] RFC 1122, Requirements for Internet Hosts- Communication Layers[S].

[16] RFC 2018, TCP Selective Acknowledgment Options[S].

[17] Dyer T D, Boppana R V. A comparison of TCP performance over three routing protocols for mobile ad hoc networks[C]// Proceedings of the 2nd ACM International Symposium on Mobile Ad Hoc Networking & Computing. Long Beach: ACM New York, 2001: 56-66.

[18] Wang F, Zhang Y. Improving TCP performance over mobile ad-hoc networks with out-of-order detection and response[C]// Proceedings of the 3rd ACM International Symposium on Mobile Ad Hoc Networking &Computing. Lausanne: ACM New York, 2002: 217-225.

[19] Sun W, Wen T, Guo Q. A novel protocol for mobile-induced packet reordering in mobile ad hoc networks[C]//Information Science and Engineering. Shanghai: IEEE Computer Society, 2008: 626-631.

[20] Ghanem T F, Elkilani W S, Hadhoud M M. Improving TCP performance over mobile ad hoc networks using an adaptive backoff response approach[C]//Proceedings of International Conference on Networking and Media Convergence. Cairo: IEEE Computer Society, 2009: 16-21.

[21] Fard M A K, Bakar K A, Karamizadeh S, et al. Improve TCP performance over mobile ad hoc network by retransmission timeout adjustment[C]//Communication Software and Networks (ICCSN), 2011 IEEE 3rd International Conference on. Xi’an, China: IEEE Computer Society, 2011: 437-441.

[22] Kim D, Toh C K, Yoo H. The impact of spurious retransmissions on TCP performance in aD hoc mobile wireless networks[C]// Personal, Indoor and Mobile Radio Communications. Athens: IEEE Computer Society, 2007: 1-5.

[23] Ibom U O. TCP performance over MANET[C]//Information Networking. Busan: IEEE Computer Society, 2008: 1-5.

[24] Goff T, Abu-ghazaleh N B, Phatak D S, et al. Preemptive routing in ad hoc networks[J]. ElsevierJournal of Parallel and Distributed Computing, 2003, 63(2): 123-140.

[25] Klemm F, Ye Z, Krishnamurthy S V, et al. Improving TCP performance in ad hoc networks using signal strength based link management[J]. Elsevier Ad Hoc Networks, 2005, 3(2): 175-191.

[26] Anantharaman V, Park S J, Sundaresan K, et al. TCP performance over mobile ad-hoc networks: Aquantitative study[J]. Wireless Communications and Mobile Computing, 2004, 4(2): 203-222.

[27] Günes M, Vlahovic D. The performance of the TCP/RCWE enhancement for ad-hoc networks[C]//Proceedings of the 7th IEEE International Symposium on Computers and Communication. Taormina, Giardini Naxos: IEEE Computer Society, 2002: 43-48.

[28] Fu Z, Greenstein B, Meng X, et al. Design and implementation of a TCP-friendly transport protocol for ad hoc wireless networks[C]//Proceedings of the 10th IEEE International Conference on Network Protocols. Paris: IEEE Computer Society, 2002: 216-225.

[29] Gajjar S, Gupta H M. Improving performance of ad hoc TCP in mobile ad hoc networks[C]//India Conference. Kanpur: IEEE Computer Society, 2008: 144-147.

[30] de Oliveira R, Braun T, Heissenb¨uttel M. An edge-based approach for improving TCP in wireless mobile ad hoc networks[C]//Proceedings of the Conference on Design, Analysis and Simulation of Distributed Systems. Orlando: The Society for Modeling and Simulation International (SCS), 2003: 172-177.

[31] Fu Z, Zerfos P, Luo H, et al. The impact of multihop wireless channel on TCP throughput and loss[C]//Proceedings of the 22nd Annual Joint Conference of the IEEE Computer and Communications Societies. San Francisco California: IEEE Computer Society, 2003: 1744-1753.

[32] Papanastasiou S, Ould-khaoua M. TCP congestion window evolution and spatial reuse in MANETs[J]. Journal of Wireless Communications and Mobile Computing, 2004, 4(6): 669-682.

[33] Nahm K, Helmy A, Kuo C C J. TCP over multihop 802.11 networks: Issues and performance enhancement[C]//Proceedings of the 6th ACM International Symposium on Mobile Ad Hoc Networking and Computing, Urbana-Champaign: ACM New York, 2005: 277-287.

[34] Zhai H, Chen X, Fang Y. Rate-based transport control for mobile ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. New Orleans: IEEE Computer Society, 2005: 2264-2269.

[35] Kliazovich D, Granelli F. Cross-layer congestion control in ad hoc wireless networks[J]. Ad Hoc Networks, 2006, 4(6): 687-708.

[36] Lien Y N, Hsiao H C. A new TCP congestion control mechanism over wireless ad hoc networks by router-assisted approach[C]// Distributed Computing Systems Workshops. Toronto: IEEE Computer Society, 2007: 84-84.

[37] Kopparty S, Krishnamurthy S, Faloutsos M, et al. Split-TCP for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. Taipei: IEEE Computer Society, 2002: 138-142.

[38] Lien Y N, Hsiao H C. Hop-by-Hop TCP over MANET[C]// Asia-Pacific Services Computing Conference. Yilan: IEEE Computer Society, 2008: 1150-1155.

[39] Giovanidis A, Stanczak S. Retransmission aware congestion control and distributed power allocation in MANETs[C]// Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks. Seoul: IEEE Computer Society, 2009: 1-10.

[40] Papandriopoulos J, Dey S, Evans J. Optimal and distributed protocols for cross-layer design of physical and transport layers in MANETs[J]. IEEE/ACM Transactions on Networking, 2008, 16(6): 1392-1405.

[41] Yu Y Q, Giannakis G B. Cross-layer congestion and contention control for wireless ad hoc networks[J]. IEEE Transactions on Wireless Communications, 2008, 7(1): 37-42.

[42] Cordeiro C D A, Das S R, Agrawal D P. COPAS: Dynamic contention-balancing to enhance the performance of TCP over multi-hop wireless networks[C]//Proceedings of the 10th International Conference on Computer Communication and Networks (IC3N). Miami: IEEE Computer Society, 2002: 382-387.

[43] Xu K, Gerla M, Qi L, et al. TCP unfairness in ad hoc wireless networks and a neighborhood RED solution[J]. Wireless Networks, 2005, 11(4): 383-399.

[44] Fu Z, Luo H, Zerfos P, et al. The impact of multihop wireless channel on TCP performance[J]. IEEE Transactions on Mobile Computing, 2005, 4(2): 209-221.

[45] Zhai H, Wang J, Fang Y. Distributed packet scheduling for multihop flows in ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. Atlanta: IEEE Computer Society, 2004: 1081-1086.

[46] Zhai H, Chen X, Fang Y. Alleviating intra-flow and inter-flow contentions for reliable service in mobile ad hoc networks[C]// Proceedings of the IEEE Military Communications Conference. Monterey: IEEE Computer Society, 2004: 1640-1646.

[47] Dyer T D, Boppana R V. A comparison of TCP performance over three routing protocols for mobile ad hoc networks[C]// Proceedings of the 2nd ACM International Symposium on Mobile Ad Hoc Networking & Computing. Long Beach: ACM New York, 2001: 56-66.

[48] Sun D, Man H. ENIC—An improved reliable transport scheme for mobile ad hoc networks[C]//Proceedings of the IEEE Global Telecommunications Conference. San Antonio: IEEE Computer Society, 2001: 2852-2856.

T. Novel delayed ACK techniques for improving TCP performance in multihop wireless networks[C]// Proceedings of the IFIP-TC6 8th International Conference on Personal Wireless Communications. Venice: Springer, 2003: 237-250." target="blank">[49] Altman E, T. Novel delayed ACK techniques for improving TCP performance in multihop wireless networks[C]// Proceedings of the IFIP-TC6 8th International Conference on Personal Wireless Communications. Venice: Springer, 2003: 237-250.

[50] Berger D, Ye Z, Sinha P, et al. TCP-friendly medium access control for ad-hoc wireless networks: Alleviating self- contention[C]//Proceedings of the 1st IEEE International Conference on Mobile Ad-hoc and Sensor Systems. Fort Lauderdale: IEEE Computer Society, 2004: 214-223.

[51] Kumaran T S, Sankaranarayanan V. Early detection congestion and control routing in manet[C]//Seventh International Conference On Wireless And Optical Communications Networks (WOCN). Colombo: IEEE Computer Society, 2010: 1-5.

[52] Yerajana R, Sarje A K. A timestamp based multipath source routing protocol for congestion control in MANET[C]//Advance Computing Conference. Patiala: IEEE Computer Society, 2009: 32-34.

[53] Shrivastava L, Tomar G S, Bhadauria S S. A survey on congestion adaptive routing protocols for mobile ad-hoc networks[J]. International Journal of Computer Theory and Engineering, 2011, 3(2): 189-196

[54] Mbarushimana C, Shahrabi A. E-TCP: Enhanced TCP for IEEE802.11e mobile ad hoc networks[C]//15th International Conference on Parallel and Distributed Systems (ICPADS). Shenzhen: IEEE Computer Society, 2009: 632-639.

[55] Mbarushimana C, Shahrabi A. Resource efficient TCP: Reducing contention-induced spurious timeouts in QoS-aware MANETs[C]//Advanced Information Networking and Applications. Okinawa: IEEE Computer Society, 2008: 1070-1077.

[56] Kazi C R, Syed F H. Explicit rate-based congestion control for multimedia streaming over mobile ad hoc networks[J]. International Journal of Electrical & Computer Sciences, 2010, 10(4): 28-40.

[57] Premalatha J, Balasubramanie P. Enhancing quality of service in MANETS by effective routing[C]//Wireless Communication and Sensor Computing. Chennai: IEEE Computer Society, 2010: 1-5.

[58] Calafate C T, Malumbres M P, Oliver J, et al. QoS support in MANETs: A modular architecture based on the IEEE 802.11e technology[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2009, 19(5): 678-692.

[59] Chaparro P A, Alcober J, Monteiro J, et al. Supporting scalable video transmission in MANETs through distributed admission control mechanisms[C]//18th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP). Pisa: IEEE Computer Society, 2010: 238-245.

[60] Chen K, Nahrstedt K. EXACT: An explicit rate-based flow control framework in MANET (extended version)[R]. Urbana: University of Illinois at Urbana-Champaign, 2002.

[61] Chen K, Nahrstedt K, Vaidya N. The utility of explicit rate-based flow control in mobile ad hoc networks[C]//Proceedings of the IEEE Wireless Communications and Networking Conference. Atlanta: IEEE Computer Society, 2004: 1921-1926.

[62] Sundaresan K, Anantharaman V, Hsieh H Y, et al. ATP: A reliable transport protocol for ad-hoc networks[C]//Proceedings of the 4th ACM International Symposium on Mobile Ad Hoc Networking &Computing. Annapolis: ACM New York, 2003: 64-75.

[63] Anastasi G, Ancillotti E, Conti M, et al. TPA: A transport protocol for ad hoc networks[C]//Proceedings of the 10th IEEE International Symposium on Computers and Communication. La Manga: IEEE Computer Society, 2005: 51-56.