【作者】 姚信威；

【导师】 王万良； 杨双华；

【作者基本信息】 浙江工业大学 ， 控制理论与控制工程， 2013， 博士

【摘要】 伴随着个人数字通信和网络社会化的快速发展,功能强大的数据终端以及多元化的网络业务,正推动着网络技术从有线向无线、从固定到移动、从文字业务向多媒体业务演化,同时网络业务的服务质量也从忽略向支持、从单一服务向多优先级区分服务进化。在这种大环境下,能实现“任何人在任何时间、任何地点以任何方式与任何人通信”的无线多媒体网络(WMN, Wireless Multimedia Network)得到了飞速的发展和广泛的应用,既能弥补传统有线多媒体网络的不足,又能提供足够的带宽满足语音、视频、图像等多媒体业务传输的需求。因此,基于服务区分的无线多媒体网络技术引起了学术界和工业界的广泛研究。本文在国家自然科学基金(61070043,61379123)和“十二五”国家科技支撑计划(2012BAD10B01)等项目的资助下,而向基于IEEE802.11e EDCA协议的无线多媒体网络,对其区分服务机制和多媒体传输性能进行深入的研究和分析,取得了如下研究成果：(1)针对无线多媒体传输建模复杂问题,首先建立基于离散Markov链的EDCA区分服务饱和模型与非饱和模型,以该模型为基础进一步提出视频流的丢包模型和时延模型,并结合视频编码结构,最终形成完整的饱和与非饱和状态下无线多媒体传输模型。根据IEEE802.11e EDCA协议的二进制退避机制,建立了基于Markov链的EDCA饱和模型,并通过该模型获取各状态的概率分布、状态间转移概率以及各优先级队列的碰撞概率、发送概率等。由于EDCA协议默认将所有的视频流数据映射至队列AC(2),综合考虑饱和状态下队列AC(2)的碰撞丢包(内部碰撞和外部碰撞)和无线信道误码丢包,建立了视频流的丢包模型；同时考虑网络延迟对于实时多媒体传输质量的影响,在EDCA饱和模型的基础上,建立了视频流的时延模型。结合上述丢包模型和时延模型,深入分析MPEG-4视频编码技术,结合其编码结构,提出实时视频传输模型,并引入可解码帧率DFR评估视频流传输质量。针对数据流的突发性和动态网络环境,在上述饱和状态下多媒体传输模型的基础上,建立更符合实际的非饱和状态多媒体传输模型。深入分析EDCA协议工作机制和后退避过程,引入特殊网络状态“空闲状态”,建立了基于Markov链的EDCA非饱和模型。同时基于该模型,分别建立视频流的丢包模型和时延模型,同时结合多媒体编码结构,提出了非饱和状态下多媒体传输模型。(2)针对EDCA协议单时隙竞争机制所导致的各优先级队列传输碰撞问题,引入了混合时隙的概念,提出了混合多时隙信道分配机制,并分别建立了饱和与非饱和状态下混合多时隙分配机制的传输性能模型。通过系统优化将三个时隙组合成一个混合时隙,各队列以混合时隙为单元进行数据传输,根据各队列的优先级别,将混合时隙中的第一个时隙固定分配至最高优先级队列AC(3)的数据传输,第二个时隙固定分配至次高优先级队列AC(2)的数据传输,而低优先级队列AC(1)和AC(0)共享第三个时隙。此外,根据基于Markov链的EDCA饱和与非饱和模型,分别建立了饱和状态与非饱和状态下混合多时隙分配机制的传输性能分析模型。实验结果和模型分析证明了所提混合多时隙分配机制显著地降低传输碰撞概北,提高各队列的吞吐量,降低各队列的传输时延,从而改善多媒体的传输质量。(3)针对EDCA协议中视频帧映射问题,利用视频编码结构特征和视频帧的重要性信息,通过跨层信息交互,分别提出了自适应前向映射机制、双重后向映射机制和拥塞感知机制,并最终融合为动态参数自适应双向映射机制。该机制在EDCA区分服务机制的基础上,充分挖掘视频分层编码的结构特征和视频帧的重要性信息,通过应用层和MAC层的信息交互,实现视频帧映射过程的跨层优化。针对不同视频帧重要性的区别和当前网络负载程度,提出了动态自适应前向映射机制和双重后向映射机制；又根据网络环境的时变特性,引入拥塞感知机制。通过与现有三种视频帧映射机制的对比分析,验证了所提参数自适应双向映射机制的优越性能。(4)针对无线多媒体网络中不同数据流业务中的服务质量问题,分别提出了基于LV生物竞争模型的无优先级区分和多优先级区分的仿生竞争模型,并采用自适应粒子群算法实现模型参数优化,在保证系统稳定性的前提下使网络资源利用率最大化。引入了Lotka-Volterra生物竞争模型,并系统地建立了生态系统与网络系统之间的映射关系,分别提出了无优先级区分的仿生竞争模型和多优先级区分的仿生竞争模型(Bio-EDCA模型)。针对Bio-EDCA模型,求解模型的稳定性、稳态输出值以及模型参数优化等问题,在实现带宽按特定比例分配的前提下最大化网络带宽利用率。在任意动态突发数据流条件下,该模型始终具有优越的稳定性和自适应性,同时保持传输友好性和避免网络拥塞。实验表明,与传统区分服务机制相比,Bio-EDCA模型不仅可以根据网络变化动态调整各优先级数据传输速率,使得网络带宽利用率达到93%(EDCA协议带宽率用率为60%-70%),而且能收敛至系统稳定状态,抗干扰能力强。更多还原

【Abstract】 With the rapid development of personal digital communication and social network, and the increasing of powerful data terminals and various network applications, which push the network technology forward from wired to wireless, from fixed to mobile, from text data to multimedia transmission, and the Quality of Service (QoS) is also promoted from be ignored to be supported, from single priority to multi-priority service differentiation. Under this condition, Wireless Multimedia Network (WMN), i.e. a network can communicate with anyone using anymode at anytime and anywhere, have been widely researched and employed in military, commercial and daily lives. WMNs can not only make up the deficiencies of tranditional wired networks, also support the transmission of multimedia applications, such as voice, video and images, etc.Based on the fundings of National Natural Science Foundation of China (Grant No.61070043,61379123) and National Science&Technology Pillar Program during the Twelfth Five-year Plan Period (Grant No.2012BAD10B01), oriented to the service differentiaon mechanism of IEEE802.11e EDCA protocol in WMN, this thesis had conducted in-depth investigation in terms of service differentiation and its performance for multimedia transmission, and has acquired considerable achievements, where the major contributions of this thesis were listed as follows:(1) According to the binary backoff mechanism of IEEE802.11e EDCA protocol, two analytical models based on Markov Chain were proposed for EDCA protocol working in saturated and unsaturated cases, respectively. The probabilities of data collision, state distribution, state transfer and access the channel of each priority queue can be derived from the proposed model. Since all the video frames are allocated to queue AC(2) according to the default rules of EDCA protocol, we built a packet loss model for video transmission by integrating the collision probability of AC(2) and wireless channel error rate. At the same time, a delay model for video transmission was also proposed by considering the effect of time delay on video transmission quality. Relying on the above packet loss model and time delay model, with the structure of MPEG-4coding, we proposed a video transmission model and introduced the Decoded Frame Rate (DFR) to evaluate the video transmission quality. Due to the characteristics of bursty data flows and dynamic network conditions, we investigated the more practical and complicated video transmission quality in unsaturated case than in saturated case. Through an in-depth study of EDCA protocol and its post backoff mechanism, an analytical model based on Markov Chain in unsaturated case was proposed by introducing a special backoff state "idle state". The corresponding packet loss model and delay model for video transmission in unsaturated case were also presented. Similarly, based on the above packet loss model and time delay model, with the structure of MPEG-4coding, we proposed a novel video transmission model for video transmission in unsaturated case.(2) Although EDCA protocol achieves service differentiation by introducing four Access Categories (ACs), its performance is deteriorated seriously due to high collision probability among ACs and underutilization of channel resources. In order to overcome these deficiencies, a novel Hybrid Slot Allocation Mechanism (HSAM) was proposed by assembling three time slots into a super slot. The main idea of HSAM is to stagger the transmission slots of different ACs to reduce their collisions. Based on the extended Markov Chain model, two comprehensive analytical models were presented to evaluate the performance of HSAM both in unsaturated case and saturated case. Extensive results from simulations have demonstrated the superior performance of HSAM than EDCA protocol in terms of throughput, packet loss and MAC delay. Moreover, the simulation results are matched very well by the derived analytical models.(3) In order to deal with the mapping problem of video frames in EDCA protocol, a Parameter Adaptive Bi-directional Mapping (PABM-EDCA) mechanism was proposed to assign each video frame with appropriate priority. This mechanism takes the structure of hierarchical video coding and the importance of each video frame into consideration, and utilizes the information of Application Layer and MAC Layer through cross-layer optimization. According to the importance of different video frames and network condition, we proposed the adaptive forward mapping mechanism and double backward mapping mechanism, respectively. Both of the above mechanisms had been equipped with congestion awareness ability. From the comparison of PABM-EDCA with other three mapping mechanisms, the experiment results showed the superior performance of PABM-EDCA under dynamic network contidions.(4) Based on the extended Lotka-Volterra (LV) biological competitive model, a bio-inspired self-adaptive rate control approach for multi-priority data transmission was proposed. This approach calculated the optimum sending rate of each flow by taking the system stability, limited available network resources and competitions from other flows into consideration. Then the proposed approach was applied to four categories of data flows defined in EDCA protocl, the allocated bandwidth of each data flow was optimized through model parameter optimization. Under the recommended parameter values, the total bandwidth utilization was maximized up to93%compared with60%-70%achieved by the EDCA protocol, whilst maintaining the service differentiation of multi-priority flows. Extensive simulation studies had been conducted to show the superior performance of the proposed approach.更多还原