【作者】 梁伟；

【导师】 张顺颐；

【作者基本信息】 南京邮电大学 ， 通信与信息系统， 2011， 博士

【摘要】 自从2005年的IEEE DySPAN学术会议上第一个认知网络概念提出以来,人们以极高的热情展开了对认知网络的研究。然而目前对认知网络探索仍处于起步阶段,认知网络体系的初步框架尚在研究探讨之中。本文尝试对认知网络的无线接入选择和拥塞控制技术进行研究。在认知网络框架构想之下,本文首先研究了无线接入选择问题。提出了基于模糊逻辑的认知信息处理架构,在跨层感知和网络感知两者上都能够实现认知过程,也可以利用各种网络设备上存储的认知信息来实现全网网络的认知功能。此外,提出了一个基于模糊逻辑决策的网络接入选择机制,这种机制允许每个用户选择能够满足QoS需求的最佳接入点。论文对认知网络拥塞控制中的相关技术进行了探讨,指出了认知网络拥塞控制研究中的主要问题,为研究的开展指明了方向。对认知网络拥塞控制传输协议进行了建模和分析,阐述了一种基于时延的端到端拥塞控制算法,称为CNFAST TCP。其特点是通过指数加权的ARMA模型来设计平滑RTT函数,达到平滑估计往返时延的目的,以消除吞吐量和排队队列在往返时延波动时产生的动荡。最后,借助控制理论中稳定裕度的概念研究了网络参数对于系统稳定性的影响。采用明确的拥塞控制算法性能评价指标进行分析,对所提出的算法进行了包括公平和稳定特性在内的多角度理论分析,系统地分析了CNFAST TCP的综合性能。更多还原

【Abstract】 Since the concept of cognitive networks was first proposed at IEEE International Symposium on DySPAN 2005, there has been launched with great enthusiasm for research on the cognitive network. However, the current exploration of the cognitive network is still in its infancy phase, the preliminary framework of cognitive network system are still under explored. In this paper, we propose a Cognitive Network approach to solve the problem of the Wireless Network Access and congestion control.Under the concept of cognitive network framework, this paper studies the problem of wireless access. A information processing framework was proposed based on Fuzzy Logic which enables the implementation of a cognition process which is both cross-layer and network-aware; furthermore, we developed the cognitive function by realizing a more complete and reliable characterization of the performance of the whole network with the cognitive information from different devices. Subsequently, we defined a network access scheme based on Fuzzy Decision Making which allows each user to choose the Access Point that best satisfies its QoS requirements.Cognitive network congestion control technologies were discussed and main related problems were pointed out. The further research trend of cognitive network congestion control was presented. With exponential weighted by ARMA models to design a smooth function of RTT, a congestion control algorithm called CNFAST TCP is elaborated in detail. It improves the smooth estimation of round-trip delay and eliminates the fluctuations of the queue throughput and queuing delay. We studied the stability margin of the network parameters on system stability. Through quantitative analysis and performance evaluation methods on the congestion control algorithms, including fairness, stability and convergence, the comprehensive performance of the high-performance transmission system is systematically analyzed.更多还原