【作者】 于洋；

【导师】 谭学治；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2015， 博士

【摘要】 随着无线通信的发展,对能够了解自身所处环境并可相应地改变其操作模式的智能无线通信系统的需求越发显著。在这一背景下,认知无线电应运而生。它代表了未来无线通信系统的一种可能的解决方案。其目的是为了提高频谱利用率,以解决当前存在的频谱利用不平衡的问题。认知无线电系统的任务包括两部分。第一部分任务为:认知无线电设备搜索并确认任何未被占用的频谱。第二部分任务是通过自适应调整发射机参数以实现通信的最佳模式。在假定空闲频谱已被检测出的前提下,本文通过设计认知无线电的智能核心——认知引擎以使认知无线电完成上述第二部分任务。本文紧紧围绕认知无线电中自适应传输问题,利用信道估计、信道分类、人工智能和链路自适应相关技术,通过研究设计信道认知引擎和认知决策引擎,以实现无线频谱资源的最优配置,从而实现通信系统高效可靠地自适应传输。首先,通过分析认知无线电中认知引擎的决策问题可被建模为多目标优化问题,因此对多目标优化理论进行了阐述和分析。在对多目标优化基本表示、凸、凹空间和Pareto最优化前沿研究的基础上,分析了以认知无线电为背景的多目标优化问题,并得出解决上述问题需用到人工智能算法和自适应传输技术这一结论。在对认知无线电中人工智能的研究基础上,重点介绍了智能优化算法。同时,在对自适应传输基本理论和物理层自适应技术的研究基础上,重点介绍了自适应调制编码技术。其次,针对用来感知外部信道状态的信道认知引擎设计问题,提出了一种既可以估计信道传递函数、又可以对信道状态进行分类的信道认知引擎。所设计的信道认知引擎通过本文提出的最小均方改进算法对信道传递函数进行估计。上述最小二乘改进算法通过设计门限以达到对传递函数的粗估计和精估计,提高了算法的估计精度。所设计的信道认知引擎通过隐马尔可夫模型和本文提出的二进制混沌粒子群算法对信道进行分类。离线状态下,采用二进制混沌粒子群算法对隐马尔可夫模型进行训练;在线状态下,采用基于隐马尔可夫模型的信道分类算法对当前信道加以分类。仿真分析表明:在山区时变信道条件下,所提出的信道认知引擎能够准确地进行信道估计和信道分类,为认知引擎作出决策提供依据。再次,针对认知OFDM系统的特点,设计了一种基于混沌粒子群算法的认知决策引擎。给出了认知OFDM的系统模型,并据此分析了认知引擎的输入量和输出量。在提出二进制混沌粒子群算法的基础上,设计了一种基于该二进制混沌粒子群算法的认知决策引擎。该引擎针对不同通信模式,通过调整权重因子,使自身沿着不同优化方向寻优,有效解决了认知OFDM系统在不同通信服务的要求下合理决策最优传输策略的问题。仿真分析表明:通过与现有其他引擎性能的比较,文中提出的认知决策引擎其优化的精确度和鲁棒性能都更优。最后,根据认知SC-FDE系统和山区时变信道的特点,设计了针对慢时变信道和快时变信道的两种认知决策引擎。针对慢时变信道,文中首先提出了一种新型自适应算法,并在此基础上,提出了基于该算法的认知决策引擎。针对快时变信道,提出了一种基于门限调整和自适应调制编码的认知决策引擎。理论和仿真分析表明:通过与认知系统中信道认知引擎、认知无线电知识库等部分地有效配合,所设计的认知决策引擎能够在山区复杂信道场景下合理决策出最优的传输策略,同时计算复杂度和工程可实现性上也具有优势,保证了SC-FDE认知系统高效高可靠性地自适应传输。更多还原

【Abstract】 With the advancements in wireless communications, there is a definite need for an intelligent wireless system that is aware of its environment and able to accordingly change its mode of operation. In this context, the cognitive radio comes into being. A cognitive-radio system represents a potential solution for future wireless communication systems. Its objective is to improve spectrum usage efficiency and minimize the problem of spectrum over-crowdedness. The operation of a cognitive-radio system is mainly divided into two tasks. In the first task, a cognitive-radio device searches and identifies any part of the spectrum that is not occupied. The second task consists of achieving an optimal mode of communication by adjusting the transmitter parameters adaptively. Under the assumption that the idle spectrum has been detected, this dissertation designs the cognitive engine, to enable the cognitive radio to complete the second task. This dissertation focuses on the adaptive transmission in cognitive radio, and studies the design of channel cognitive engine and cognitive decision engine based on channel estimation, channel classification, artificial intelligence and link adaptation techniques to achieve the optimal configuration of radio spectrum resource and efficient, reliable adaptive transmission.Firstly, the decision process of cognitive decision engine in cognitive radio can be modeled as a multi-objective optimization problem. Thus the multi-objective optimization theory are described and analyzed. This dissertation analyzes the multi-objective optimization issue in the context of cognitive radio based on studying the basic representation for multi-objective optimization, convex space, concave space and Pareto optimization edge. Then it can be concluded that the artificial intelligence algorithms and adaptive transmission techniques need to be used to resolve the above mentioned multi-objective optimization issue. The adaptive modulation and coding technique is focused on based on the study of the basic theory of adaptive transmission and physical layer adaptive techniques.Secondly, a novel channel cognitive engine is proposed to sense the external channel state, which can not only estimate channel transfer function, but also classify the channel state. The improved least squares algorithm is used by this engine for channel estimate. The above mentioned algorithm can improve the precision by designing the threshold and processing the coarse estimate and fine estimate of the channel transfer function. The channel classification is processed by hidden Markov model and binary chaotic particle swarm optimization algorithm given by this dissertation. The hidden Markov model is trained by the binary chaotic particle swarm optimization algorithm offline, while the hidden Markov model based algorithm is used for channel classification online. Simulation results and analysis show that under the conditions of the time-varying channels in mountainous area, the proposed channel cognitive engine can accurately estimate and classify the channel and provide the basis for the cognitive engine to make decision.Thirdly, According to the characteristics of the cognitive OFDM system, a chaotic particle swarm optimization algorithm based decision engine is designed. The model of cognitive OFDM system is given, and then the input and output of the cognitive engine are analyzed. This dissertation proposes a binary chaotic particle swarm algorithm. Then a cognitive decision engine based on it is designed. According to different communication modes, this engine can adjust the weighting factors for optimization along different optimization directions, which can resolve the issue of deciding the optimal transmission scheme effectively with the requirements of different communication services in cognitive OFDM systems. Simulation results show that the proposed cognitive decision engine, which has higher fitness value and stronger robustness, is better than the other existing engines.Finally, according to the characteristics of the SC-FDE system and the time-varying channels in mountainous area, two cognitive decision engines are designed for slow time-varying channels and fast time-varying channels respectively. This dissertation presents a new adaptive algorithm and then proposes a decision engine based on the above algorithm for slow time-varying channels. Moreover, a threshold adjustment and adaptive modulation and coding based cognitive decision engine for fast time-varying channels. Theory and simulation analysis show that the proposed cognitive decision engines can decide the optimal transmission scheme reasonably under the complex conditions of the time-varying channels in mountainous area cooperated with channel cognitive engine and cognitive radio knowledge base, while they have an advantage in terms of the computational complexity and project realization. Thus, the proposed cognitive decision engines ensure the SC-FDE cognitive systems with high reliability and efficient adaptive transmission.更多还原