节点文献

基于数据接入类型和节点负载的移动自组织网络QoS按需路由协议

A QoS Support On-demand Routing Protocol Based on Access Category and Node Load in Mobile Ad Hoc Networks

【作者】 于永娇

【导师】 李月;

【作者基本信息】 吉林大学 , 信号与信息处理, 2011, 硕士

【摘要】 移动自组织网络(Wireless Mobile MANET Network),亦称无线自组织网络。由于移动自组织网络自身灵活组网、成本低廉等独特优点,被广泛应用于军事、医疗、灾难援助和电子商务等领域。目前,越来越多基于UDP的多媒体业务数据在无线网络中传输。首先,这些多媒体数据常常具有流量大、持续时间长的特点,这就加重了网络负担,增大了发生网络拥塞的概率;其次,当UDP数据在传输过程中遇到网络拥塞时,它不具备相应的退避机制,这导致基于UDP传输的数据比基于TCP传输的数据消耗更大的网络带宽。近些年,为了对移动自组织网络传输多媒体数据提供QoS支持,学者们提出了很多对传统移动自组织网络路由协议的改进。Prof. Jitendranath在原有的AODV路由协议提出添加QoS标志位和两个16bit字段(Cost和Delay),在转发请求之前需要更新Cost字段和Delay字段(如果需要延时保障),路由请求节点选择消耗最小的路由以达到负载均衡的目的。Yuan提出对节点负载设定一个门限值,通过依据节点负载是否超过门限值选路以减轻网络拥塞,从而实现提供QoS的目的。然而现有能够提供QoS保障的移动自组织网络路由算法有的仅局限在路由层,有的利用了端到端时延、节点负载和节点能量等信息。IEEE的802.11e工作组为了提供QoS支持,定义了四种不同的数据接入类型。首先,不同接入类型(AC,Access Category)使用不同的仲裁帧间间隔(AIFS,Arbitration Interframe Space)以获得不同的信道接入机会,对于AIFSN值较小的情况,优先级高的数据接入类型相对于优先级低的接入类型会更早的减小他们的退避计时器。其次,EDCA对不同的AC在退避过程中分配不同的竞争窗口最大值和最小值( CWm in和CWm ax),而退避间隔在[0, CWi ]间随机选择(其中CWi = 2ik?1 CWmin, k为退避级别),对于高优先级的接入类别会拥有较小的CWm in和CWm in。本文提出的QCAODV路由协议使用一种简单而有效的方法实现移动自组织网络的QoS保障。与现有的利用端到端时延信息、节点负载信息等路由算法不同,我们采用跨层QoS结构体系思想,提出考虑数据接入类型对网络状态的影响,MAC层和路由层相辅相成实现移动自组织网络的QoS保障和网络拥塞避免。首先,我们的QCAODV是工作在与MAC层连接处的跨层解决方案,因此,任何时候节点转发数据的接入类型发生变化都由MAC层负责向IP层更新信息。在每次路由发现过程中,我们提取数据帧首部中包含的针对不同数据类型的不同优先级信息对所使用路由上的节点做标记,使用考虑了服务类型信息影响后的路由使用度量(Metric)来判定最佳路由。这样,路由请求发起节点收到多个RREP时,选择Metric值最小的路径作为最佳路由。其次,在AODV路由协议中,当中间节点收到路由请求节点发出的RREQ时,如果路由表中有通往目的节点的路由,允许中间节点应答。中间节点应答能够减少RREQ泛洪,但是容易引起重叠路由而产生拥塞。我们在QCAODV中规定中间节点不能发出RREP,这样做能够利用最新的节点负载数据接入类型信息判断最佳路由,减少了由于不同优先级数据争用发送机会而引起网络拥塞的概率。本文考虑节点负载数据类型对网络状态的影响,应用了802.11e对QoS保障的支持,在AODV路由协议的基础上提出了QCAODV路由协议。由于在路由发现过程中,对活动节点缓冲队列中的数据接入类型进行监控,并作为最佳路由判定的因素。有效降低了产生重叠路由的概率,从而降低了由此产生的时延增大、丢包率上升和网络吞吐量降低等问题。通过仿真实验证明,QCAODV能够为移动自组织网络提供较好的QoS保障。

【Abstract】 The Ad Hoc network is also called MANET. Because of the flexible networking, low cost, and some other unique advantages of Ad Hoc, it has been widely used in military, disaster relief and e-commerce, etc. Nowadays, more and more multimedia services data , such as VoIP, video conference and so on, which are based on UDP protocol , are transmitted by Ad Hoc networks. First of all, the multimedia data are often large flow in long duration, which increase the network burden and the rate of congestion. Secondly, UDP has no back off algorithm when it meets congestion during the transmission, so UDP based data consumes more bandwidth than TCP based data.In recent years, in order to provide QoS support for Ad Hoc, scholars have proposed many improvements to traditional Ad Hoc routing protocols. Prof. Jitendranath proposed to expend two 16 bits fields(Cost and Delay) and building up a QoS sign bit in RREP (route reply) and RREQ (route request) respectively. Intermediate nodes update the Cost and Delay (if delay sensitive) before forward the RREQ, then the source node chose the route which has smallest cost to achieve load balance. Yuan uses a threshold to illustrate loading state and filter paths based on these informations to achieve QoS support. But the existing routing protocols with QoS guarantees often are limited in routing layer and make use of the informations related to end-to-end delay , node loads or node power. None of them put forward exploring the node data access categories information.The 802.11e work group of IEEE defined 4 data access categories (ACs) for transmitting frame with priority. First of all, each ACi assigned different arbitration inter-frame space (AIFS) to obtain corresponding channel access opportunity. If the value of AIFSN is smaller, the ACi with high priority will start to reduce its back off counter earlier than the ACi with low priority. Moreover, In EDCA mechanism, each ACi has different minimum and maximum contention window size during the back off period. The access categories which have high priorities are assigned small value of CWmin and CWmax. Then the back off interval time is chosen in the range [0, CWi] randomly, where CWi=2ik-1CWmin and k illustrate the back off stage.We propose QCAODV routing protocol which is simple yet effective to gain QoS support for Ad Hoc. Contrast to those existing routing protocols using end-to-end delay, node load or node power information, our protocol make the use of cross layer QoS structure system idea, consider the effect of data access categories to network state, implement QoS guarantees and congestion avoidance in both MAC layer and routing layer.First of all, QCAODV is a cross layer routing solution which works in the conjunction between MAC layer and routing layer. Therefore, MAC layer is responsible for updating node data category changing information to IP layer. During the route discovery process, we detect the priority information contained in the frame head and mark the nodes on the path. Then we utilize the new routing metric considering the effect of access categories to choose the optimal path. Then route request nodes will choose the path with smallest metric as route when they receive multiple RREPs.Secondly, in AODV, intermediate nodes will send RREP if they have the path to destination when they receive RREQ. Intermediate nodes reply could reduce RREQ flood but it is also generate overlap routes, thus lead to congestion. In QCAODV,we prescribe that intermediate nodes are prohibited to reply even they have the routes to the destination. In this way, the latest node data access category information could be used for route judgments and decrease the possibility of network congestion due to contention of transmission opportunity between different ACs.In this paper, considering the effect of node load data access categories on network state, we proposed QCAODV routing protocol based on original AODV. During the route discovery process, mobile nodes monitor the data type buffered in its queue and utilize this information to choose optimal route. Our protocol decreases the probability of different ACs use the same node to transport data, thereby improves the performance parameters of Ad Hoc network. Through the simulation experiments, QCAODV provides better QoS guarantees.

  • 【网络出版投稿人】 吉林大学
  • 【网络出版年期】2011年 09期
  • 【分类号】TN929.5
  • 【被引频次】1
  • 【下载频次】60
节点文献中: 

本文链接的文献网络图示:

本文的引文网络