【作者】 张国翊；

【导师】 张平； 胡铮；

【作者基本信息】 北京邮电大学 ， 电路与系统， 2011， 博士

【摘要】 随着信息服务高带宽化,内容形式多元化,提供方式智能化的需求不断增长,无线网络正朝着更高传输速率、更广覆盖范围、融合多种异构接入网络,并与多种周边网络高效协同的方向发展,但仍然面临着诸如网络资源尤其是频谱资源高效利用、网络高能效数据传输以及网络认知性和适变性等关键技术挑战。一方面,随着基于频谱的无线业务和终端设备的显著增加,人们对无线电通信频谱资源的需求越来越大,“频谱资源稀缺”已经成为制约下一代无线通信系统发展的主要因素之一。在此情况下,如何对网络资源尤其是频谱资源实现高效动态分配与管理已成为无线网络亟待解决的问题。另一方面,考虑到无线网络中的终端用户能量有限,除了资源的高效利用以外,如何在满足自身数据传输可靠性的前提下,有效降低整个系统的能量消耗,延长网络的生存时间,这也是目前无线网络在节能方面所面临的挑战。此外,面对复杂多变的业务需求和瞬息万变的外部环境,迫切需要突破原有无线网络设计思想,为网络引入认知能力和学习能力,并通过与外界环境信息交互,设计出具有环境感知功能的异构融合网络体系结构及相应的感知推理机制,从而解决静态网络模式与动态需求之间的矛盾。为了有效解决上述问题,论文一方面着眼于认知无线电技术在提供频谱资源高效利用和无线网络适变性方面的长处,研究并探讨了动态频谱管理机制和用户环境感知推理技术,另一方面则重点关注于协作通信技术在提高系统抗衰落性能、信道容量及传输可靠性方面的优势,并在此基础上将认知无线电技术和协作通信技术相结合,进行了涵盖信道分配与功率控制的无线资源联合优化方面的研究工作,主要贡献如下：针对多个授权用户和认知用户共存的无线网络环境,设计了多阶段博弈框架以联合描述阶段内频谱分配和阶段间频谱交易这两个密不可分的过程。在此框架下,一方面从最大化系统效用角度出发,提出了阶段内基于多用户频谱共享机制的改进K-M算法来对频谱资源进行优化分配；另一方面,基于前一阶段内的频谱分配结果,重点关注阶段间频谱交易过程中的用户间动态交互行为以及以用户为中心的策略自适应调整机制,并从用户局部利益出发,分别提出了基于进化博弈的认知用户频谱选择算法以及基于非合作博弈授权用户策略调整机制,并给出了该混合博弈模型的均衡解,从而在满足用户自身需求的同时保证不同区域(群组)用户间的公平性。最后,结合频谱交易模型实例,分别对认知用户进化均衡和授权用户纳什均衡的存在性和稳定性进行了分析和阐述,并通过大量仿真揭示了该混合博弈模型及其均衡状态在不同系统参数设置下的网络动态性和自适应性,以及用户策略调整行为对于系统性能的影响。针对认知无线网络多终端协作通信场景,在源节点和目的节点之间实现了基于多种传输方式(直接传输、复用(双跳)传输和中继传输)的并行数据传输,并在此基础上,针对多种传输方式并存条件下的不同信道分配模式,分别提出了基于最大化系统传输容量和基于最大化网络生存时间的信道分配与功率控制联合优化方案,从而达到提高系统传输容量,有效降低系统能耗及延长网络生存时间的目的。最后,通过仿真比较了不同信道分配模式在系统传输容量方面的性能,同时也分析了任一信道分配模式下,并行传输方式较单一传输方式在系统能耗成本及网络生存时间这两方面的优越性。更重要的是,通过引入能量价格激励机制,阐述了在相同并行传输方式条件下,传输速率分配比例以及能量价格因素分别对系统能耗成本及网络生存时间的影响。针对认知中继网络与授权网络共存的应用场景,在充分考虑认知用户在各个信道上的不同传输功率以及认知用户对授权用户出现概率的误检或者漏检的情况下,分别对认知无线网络协作传输系统中频谱感知和数据传输这两阶段进行了深入分析,并在此基础上提出了基于最大化系统传输能效的感知传输时间分配和功率控制联合优化方案,以达到提高单位能量的传输比特数的目的。通过仿真,比较说明了协作传输模式较非协作传输模式在系统传输能效方面的优越性,且随着感知传输时间分配比率的减小,系统传输能效性能将得到进一步提高。此外,考虑到用户频谱感知和数据传输时间均严格受限于其帧长度,证明了采用顺序优化的方式能够有效地获得该联合优化方案的全局最优解。针对网络认知性和适变性问题,结合异构融合网络环境下的上下文特点,首先分析了用户环境感知的内涵,并在构建异构融合网络用户环境感知总体框架的基础上,设计了用户环境感知功能实体及其交互机制,并着重阐述了用户环境感知推理流程。此外,针对感知推理机制中常见的BP神经网络模型收敛速度慢且易陷入局部极小的缺陷,提出了一种基于构造复合误差函数和分层动态调整不同学习率的BP改进算法,并采用Lyapunov稳定性原理分析了改进算法的收敛性,同时对改进算法的具体流程进行了描述。研究分析表明,该BP改进算法能够有效地克服传统BP神经网络模型的缺陷且算法稳定收敛,是一种行之有效的用户环境感知推理方法。论文最后对全文进行了总结,并指出了今后的研究方向。更多还原

【Abstract】 With the growing demands for high bandwidth services, diversity in content forms, intelligent ways for information service provision, future wireless network is developing rapidly with the features of higher transmission rate, broader coverage, integration of heterogeneous access networks and effective coordination with various perimeter networks. However, key technical challenges still exist in terms of efficient network resource usage, especially the efficient use of spectrum resources, the network energy-efficiency of data transmission and the cognition and adaptation to the network variability.On the one hand, with a significant increase of the spectrum-based wireless services and terminal equipment, the demand for spectrum resources is growing rapidly, which leads to the "spectrum scarcity" becoming one of the main factors for constraining the development of next generation wireless communication systems. In this case, how to allocate network resources, especially for efficient dynamic spectrum allocation and management of wireless networks, has become an urgent problem to be solved. On the other hand, considering the end-users in wireless network has limited energy, in addition to the efficient resource usage, how to effectively reduce the energy consumption of the whole system as well as prolonging the network lifetime under the premise of satisfying users’ own reliable data transmission, which is also the challenge of wireless network faced currently in terms of energy conservation. Moreover, facing to the complex and changeable service requirements and external environment, it’s urgent need to break through the existing wireless network design by introducing cognitive and learning ability, and through exchanging information with the environment, it is also required to design the heterogeneous converged network architecture with context-aware features as well as the corresponding user context-aware reasoning mechanism to solve the contradiction between static network mode and dynamic demands.In order to effectively address above problems and challenges, on the one hand, by focusing on the merits of cognitive radio technology in terms of efficient spectrum resources usage and wireless network adaptation, this dissertation focus research on the dynamic spectrum management mechanism and user context-aware reasoning technique. On the other hand, on the basis of taking advantages of cooperative communication technology in terms of the improvement of system combating fading performance, channel capacity and transmission reliability, this dissertation mainly combines above two technologies and investigates the joint optimization of radio resources incluing channel allocation and power control. The main contributions of this dissertation include following aspects:Considering a wireless network environment in which multiple primary users and secondary users coexist, a multistage game-theoretic framework is proposed for jointly modeling the inseparable intra-stage spectrum allocation and inter-stage spectrum trading process. In this framework, on the one hand, the improved K-M algorithm based on multi-user sharing mechanism is presented in intra-stage spectrum allocation from the perspective of system utility maximization. On the other hand, according to the spectrum allocation results within the previous stage, the multiuser interactions and user-centric strategy adaptive adjustment mechanism are mainly focused on the inter-stage spectrum trading. Also, from the perspective of individual payoffs, the spectrum-selection algorithm based on evolutionary game for secondary users and strategy adjustment mechanism based on non-cooperative game for primary users are proposed respectively, and the equilibrium solutions of above hybrid game model are obtained, which can not only satisfy users’own needs but also guarantee the fairness among users on different region (group). Finally, for a specific spectrum trading exemplification, the existence and stability of evolutionary equilibrium for secondary user and Nash equilibrium for primary user are analyzed and elaborated, respectively. A number of simulation results reveal the network dynamics and adaptation of hybrid game model and their equilibrium status under different systems parameters, and investigate the effect of user strategy adjustment behavior on system performance.Considering a multi-terminal cooperative communication scenario in cognitive wireless networks, the dissertation implements the parallel data transmission based on various transmission modes (direct transmission, multiplexing (dual-hop) transmission, and relay transmission). In addition, on the basis of various types of transmission coexistence, the joint optimization scheme for channel allocation and power control based on transmission capacity and network lifetime maximization are proposed respectively under different channel allocation models, which has the purpose of improving system transmission capacity, reducing energy consumption and extending the network lifetime. Simulations compare the performance of system transmission capacity among different channel allocation modes and analysis the superiority of parallel transmission over single transmission in terms of system energy-consumption cost and network lifetime simultaneously. More importantly, by introducing the energy price incentive mechanism, under the condition of the same parallel transmission, the dissertation also elaborates the influence of the factors that transmission rate allocation proportion and energy price on system energy-consumption cost and network lifetime respectively.As for the cognitive relay network and licensed network coexistence scenario, taking full account of different cognitive user’s transmission power over each channel and their mis-detection or false alarm probability of licensed users, the dissertation makes in-depth analysis of the spectrum sensing and data transmission phase in the cooperative transmission system, and on this basis, the joint optimization scheme for sensing-transmission time allocation and power control is proposed to improve the transmission bits of per unit of energy. Simulation results show the superiority of relay-assisted transmission mode over non-relay transmission mode in terms of system transmission energy-efficiency, and with sensing-transmission time allocation ratio decreasing, system performance of energy-efficiency can be further improved. Moreover, observing time durations of both spectrum sensing and data transmission are within a strict interval, it is proved that optimal strategy of sensing-transmission time and power allocation can be tractable by sequential optimization effectively.As for the cognition and adaptation of network issue in heterogeneous converged network environment, the dissertation describes the characteristics of context in the converged network environment and analysis the content of user context-awareness. Based on establishing the heterogeneous converged network architecture with user context-aware features, user context-aware functional entities and their interaction mechanism are designed, and the corresponding context-aware reasoning process is also elaborated. Moreover, as for the common reasoning mechanism like BP neural network which has the shortcomings such as low convergence rate and easy plunging into local minimum, an improved BP algorithm is proposed with construction of the composite error function and dynamical adjustment of different learning rates, also the convergence of the improved algorithm is analyzed based on the principle of Lyapunov stability and the algorithm process is described simultaneously. Theoretical analysis shows that the improved BP algorithm can overcome the shortcomings of traditional BP neural network effectively and has the property of stable convergence, so it is an effective method for the user context-awareness and reasoning.A summary is given at the end, where the future research directions related to this doctoral dissertation are also pointed out.更多还原