【作者】 陈远；

【导师】 李乐民；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2005， 博士

【摘要】 随着无线网络的发展,移动通信用户数和Internet用户数急剧增加,人们期望新一代移动通信系统不仅具有更大的容量,还要支持移动多媒体业务,除了提供话音业务外,还支持低/高速数据、图像等非话音业务的传输。不同业务有不同的服务质量(QoS)要求,如对时延、分组丢失率、数据速率的要求不同。无线网络设计有两大目标:一是保证各类业务的QoS要求,二是使网络的资源利用率达到最大,这需要借助于无线资源管理。第三代移动通信系统的无线资源管理主要包括呼叫接纳控制、切换控制、功率控制、负荷控制、分组调度等。本文重点研究了多媒体无线分组网络中的缓存管理与分组调度算法。 在绪论部分,对无线网络的发展做了简单的介绍,对缓存管理和分组调度算法的作用及现有算法进行概述,并提出存在的问题。 缓存管理与分组调度同样需要考虑信道条件的影响,第二章提出了一种兼顾信道条件和公平性的无线分组丢弃算法,根据缓存的拥塞情况自适应的在信道条件和公平性之间取得平衡。当发生轻微拥塞时,算法将倾向于公平的丢弃分组,以保证所有用户按比例的获取系统资源;当发生中度拥塞时,算法将倾向于丢弃信道条件较差的业务流的分组,结合调度算法暂时降低其发送速率以缓解拥塞;当发生重度拥塞时,算法又将倾向于公平的丢弃分组,通过上层的流控机制,在满足QoS要求的前提下,公平地降低大部分业务流的发送速率,以加速缓解拥塞。 为了改善网络中RED算法的公平性和自适应能力,在第三章引入新的变量——分享指数,取代单个业务流缓存占用量来表征不同业务流对网络资源的占用情况,以克服缓存占用量表征公平性的缺陷。同时,在无线网络中对分享指数的定义进行推广,使其同时兼顾公平性和系统性能(信道条件)。通过对分享指数和队列长度共同制定规则,提出了一种基于模糊逻辑控制的缓存管理算法。算法在根据缓存队列长度计算丢弃概率时,根据不同流对网络资源的占用情况(即分享指数)进一步决定增加或减小更多还原

【Abstract】 With the development of mobile communications, mobile users and Internet users are increasing dramatically. People expect that next generation mobile communication systems can provide larger capacity and support mobile multimedia services. Besides providing real-time speech service, next generation mobile communication systems are required to support other services such as low/high rate data, pictures etc. Heterogeneous services have different quality of service (QoS) requirements, for example, the requirements of time delay, packet loss rate, and transmitting rate for heterogeneous services are differentiated. There are two main objects for wireless network design, one is to guarantee QoS requirements of heterogeneous services, and the other is to make resource utilization maximized. This recurs to radio resource management (RRM). Radio resource management for the third generation mobile communication systems includes call admission control, handoff control, power control, load control and packet scheduling. This dissertation pays attention to buffer management and packet scheduling in wireless multimedia packet networks.A brief introduction on the evolution of wireless networks is presented in chapter 1, followed with an overview on buffer management and packet scheduling algorithms. We also give some advices and basic thoughts on these problems.As well as packet scheduling algorithms, buffer management algorithms also need to consider channel condition. In chapter 2, a fair packet dropping algorithm is proposed to decide dropping policy in wireless networks when congestion occurs. The algorithm considers both channel condition and fairness so as to achieve tradeoff between throughput and fair services, when slight congestion occurs, algorithm trends to drop packet fairly, so as to grant each flow acquire the resource according their proportion; when moderatecongestion occurs, algorithm trends to drop packet of flows which has worse channel condition, to reduce send rate of these flows, so as to relieve congestion; when severe congestion occurs, algorithm trends to drop packet fairly again, through flow control mechanism of upper layer, sending rate of all flow will be greatly reduced, so as to accelerate congestion relief.In order to improve the fairness and adaptability of RED algorithm, the buffer occupancy of individual flow is replaced by a new introduced variable -Sharing Index to indicate the network resource share among difference flows, so as to overcome the limitation of using buffer occupancy. The definition of Sharing Index in wireless networks is also extended, to make it consider both fairness and system performance (channel condition). Through defining rules of both Sharing Index and queue length, a fuzzy logic based buffer management algorithm named FF-RED is proposed. When calculating packet dropping probability in FF-RED algorithm, the probability will further increase or decrease based on the sharing of network resource (by means of Sharing Index), so it can use the difference between flows to make congestion relief more quickly, and it has better adaptability and fairness.In chapter 4, a random early expiration detection based buffer management algorithm for real-time traffic over wireless networks is propsed. The main idea is to predict whether new arrival packet can reach the receiver under the limit of maximal delay bound, and random discard the packet based on estimate result, so as to prevent invalid transmission wasting wireless resources, and at meanwhile to decrease queue delay and to reduce expiration probability of following packets. Considering the characters of wireless link such as variable channel condition and transmission error, the algorithm is composed of three steps: aimed to guarantee the delay of retransmission packets, using the method which is similar to early congestion detection to detect expiration and discard packets in advance; adaptive adjusting the drop probability based on the trend of channel condition; re-calculating the threshold value when the link rate is changed. The influence on the TCP更多还原