【作者】 马安生；

【导师】 房至一；

【作者基本信息】 吉林大学 ， 计算机系统结构， 2008， 硕士

【摘要】 随着局域网和Internet的快速发展,宽带无线接入在“最后一公里”网络接入中得到了广泛的应用,然而宽带无线接入要比有线接入受更多问题的制约,如无线带宽资源有限、传输容易受到干扰、较低的链路可靠性、设备电池能量问题和移动性的切换问题,为了克服这些问题产生的影响,必须对宽带无线接入网络提供QoS保证机制。IEEE802.16标准是一种新兴的无线城域网宽带接入技术,它突出的特点就是在MAC层提供灵活的QoS机制,可以为用户提供多种业务类型QoS保证,但在标准中并没有定义接入控制和业务调度算法的具体实现,因此在接入控制策略和调度算法方面的研究成为热点。本文通过对IEEE802.16标准MAC层QoS保证机制研究分析,提出了基于预留最小带宽的接入控制策略和符合业务类型的多级调度算法。并在Linux操作系统下建立基于NS2的系统仿真模拟,使用Tcl脚本程序语言和C++语言实现对IEEE802.16标准MAC层QoS保证机制的模拟,最后采用Gawk程序语言对模拟产生的大量数据进行自动化分析,分析结果证明我们提出的解决方案能够很好地保证多种业务类型的QoS。更多还原

【Abstract】 With the rapid development of LAN and Internet, the higherrequirement of communication network has been put forward. Broadbandcommunication network is the inevitable development trend. At present thefiber as the primary medium for the backbone network has met thebandwidth demand basically. The main bottleneck lies on the“lastkilometer”, namely the bottleneck exists between the user and the backbonenetwork. In order to solve the issue, as an effective way, the terminalconnection may be replaced with broadband wireless access.IEEE proposes the IEEE802.16 standard, which provides a globalunified standard for broadband wireless access. However, compared withthe wire transmission, there are more constraints of the wirelesstransmission such as the limited wireless bandwidth, the susceptibletransmission, the low link reliability, the equipment battery energy problemand mobile handoff problem, therefore, many types of services in thenetwork transmission need the essential quality of service (QoS) support.The prominent characteristic of the IEEE802.16 standard is to provide aflexible QoS mechanism in the MAC layer, which includes two main parts,one is the management mechanism such as the division of the schedulingservice type, the dynamic service flow management, the polling, thesignaling architecture of broadband request and distribution, these aredefined in the IEEE802.16 standard on detailed; and the other hand is thecorresponding guarantee mechanisms, which includes admission control,scheduling algorithm, buffer area management and so on. Unfortunately,these guarantee mechanisms have not carried on the definition and theelaboration in the standard. Whereas good admission control mechanismand efficient scheduling algorithm can satisfy the users’QoS requirements, optimize the limited wireless resources, make full use of the limited wirelessfrequency spectrum resources, and improve the wireless network broadbandutilization. With the above advantages not only dose it provide more userswith better service, but also maximize the benefit of service provider.Therefore, how to design an excellent guarantee mechanism becomes theresearch focus in recent yearsThe paper gives the specific description and analysis about the QoSguarantee mechanisms in the IEEE802.16 standard MAC layer. On the basisof the analysis, we propose admission control strategy based on theminimum reserved bandwidth and the multi-level scheduling algorithmaccording with the service type.The IEEE802.16d standard defines four QoS scheduling service types,which are unsolicited grant service (UGS), real-time polling service (rtPS),non-real-time polling service (nrtPS) and best effort (BE) service. A newservice called extended real-time polling service (ertPS) is added to theIEEE802.16e standard. Different scheduling types bearing different services.The UGS is responsible for T1/E1 and VoIP service with no silencesuppression. The ertPS bears VoIP service with silence suppression. ThertPS bears MPEG video service. The nrtPS bears FTP service. The BEservice provides short message and e-mail services. Considering with thedifferent QoS requirements by the five service types, the paper purposes anew admission control strategy based on the minimum reserved bandwidthwhich conforms to the IEEE802.16 standard. We divided the totalbandwidth into five parts in this kind of admission control strategy. Thedivided parts are used for minimum reserved bandwidth of UGS, ertPS, rtPS,nrtPS and the public reserved bandwidth. The sizes of minimum reservedbandwidth of UGS, ertPS, rtP, nrtPS are determined by the situation of thehistory network load. The BE service takes the public reserved bandwidthdirectly. When a new service applies for access, it will be directly accessed if the minimun reserved bandwidth can satisfy its need. Otherwise, with theservice QoS guaranteed, the accessed services, including ertPS, rtPS, nrtPS,in the corresponding minimum reserved bandwidth will be degraded inorder to save out part of bandwidth for the new service. The degradation isdefined that the current transmission rate is decreased to the minimumtransmission rate with meeting QoS requirement. With the measurement, theservice may access network if the saved bandwidth contents the demand,otherwise the service will directly occupy the public reserved bandwidth.The UGS doesn’t degrade the accessed service because UGS is the speedinvariable service. It directly occupies the public reversed bandwidth. Thenjudge if the reversed bandwidth is enough. If the public reversed bandwidthis not adequate it will reduce the transmission speed of BE service till tozero. Apart from BE service, all services have the minimum reversedbandwidth, so each one can get service when the network load is heavy.In the aspect of scheduling algorithm, we bring forward a multi-levelscheduling algorithm according to the service type. The multi-levelscheduling algorithm mainly considers two aspects. On the one hand, itshould guarantee the fairness of all types of service. Namely, the servicewith low priority won’t be starved although there are a great mount ofservices with high priority in the network. On the other hand, it shouldensure the real time requirements are given more services. Based on the twoconsiderations, the IWFQ, a kind wireless scheduling algorithm, is adoptedas the first-class scheduling algorithm. The IWFQ scheduling algorithm notonly guarantees the fairness of the wireless, but it also considers the actualtransmission of wireless link and provides compensation for the laggedstream and provides penalties for the advanced stream. As the second-classschedule, different types of services should adopt the different schedulingalgorithms. Since the BS allocates bandwidth in real time for the serviceperiodically in UGS and ertPS, the periodic scheduling strategy is adopted. After the first-class schedule, Weighted Round-Robin scheduling algorithmshould be used in the second-class for rtPS, nrtPS and BE service. We willset different priority weights according to the different priority of rtPS,nrtPS and BE service, namely the service with high real-time requirement,such as rtPS, should be set higher weight so as to guarantee the more pollingtimes, and the opposition, like nrtPS and BE, should be set a lower weight.As the extreme case, the weights of UGS and ertPS will be assigned thebiggest value as it allocates bandwidth with fixed period and demandshighest real-time. As the minimum bandwidth is reserved for the rtPS andnrtPS, the starving phenomenon can be avoided, even though there are manyservices of UGS and ertPS in network.In the last part of this paper, we established the simulation with NS2 onLinux operating system, realized the simulation of QoS guaranteemechanisms in the IEEE802.16 MAC layer. Using NAM showed the realtime simulation results. Using Gawk programming language analysed theamounts of data in the trace file and using Gnuplot showed the visual mapand analysed the results. The experiment indicates the method we providecan guarantee the QoS of various services well.更多还原