认知无线电网络中的频谱预测技术研究

发布时间：2018-06-25 10:03

本文选题：认知无线电 + 频谱预测　；参考：《北京交通大学》2014年博士论文

【摘要】：频谱感知、频谱决策、频谱共享、频谱搬移是认知无线电技术的四大功能。次级用户通过频谱感知发掘频谱空穴,并通过频谱决策、频谱共享、频谱搬移三个功能对发掘的频谱空穴加以利用。传统的频谱感知、频谱决策、频谱共享和频谱搬移通常造成较大的处理时延、反应时延、以及能量消耗。为了解决上述问题,研究人员提出了频谱预测技术,利用不同的方法实现对未来时刻的频谱状态的预测。已有工作尚未针对如何利用次级用户之间的合作,改善未来时刻频谱状态预测准确率这一问题展开研究。此外,如何预测更多的频谱参数,并充分利用这些参数进一步改善认知无线电网络的整体性能,也是频谱预测技术研究领域有待解决的难点之一。本论文针对上述难点问题,对认知无线电网络中的频谱预测技术进行深入详细的研究。主要内容及创新点包括： 1.研究了认知无线电网络中的协作频谱状态预测问题。认识到已有的单个次级用户单独进行的本地频谱状态预测方法预测准确率有限的问题,充分利用次级用户之间自发的合作,提出一种新的协作频谱状态预测方法,改善频谱状态预测的准确率。具体而言： 1)将次级用户之间的合作过程建模为一个合作博弈过程,并基于合作博弈理论以最大化次级用户的频谱预测准确率为目标提出了一种新的合作组形成算法。每个合作组内选择一个次级用户担任首领,负责汇总组内其他用户的本地频谱状态预测结果,并通过数据融合获得协作频谱状态预测结果。 2)将设计的协作频谱状态预测方法扩展到任意多个初级用户任意多个次级用户构成的认知无线电网络中,改善算法的普适性。基于任务分配的思想,将次级用户分类,每一类用户只对一个初级用户的授权频谱进行协作频谱状态预测,有效降低了多个初级用户场景下协作频谱状态预测的复杂度。 2.研究了基于频谱状态持续时间预测的次级用户频谱感知间隔优化方法。传统的频谱感知技术要求次级用户在每个时隙的开始阶段进行一次频谱感知。而真实频谱使用数据则表明任一频谱状态都以一定概率持续若干时隙。在上述背景下,次级用户的最优频谱感知间隔,即次级用户两次频谱感知之间间隔多少个时隙,成为一个值得研究的课题。具体的研究包括： 1)次级用户用一个隐马尔科夫模型来描述频谱状态的变化规律,并通过分析该模型预测频谱状态的持续时间。 2)基于对频谱状态持续时间的预测结果,进一步预测次级用户采用某一频谱感知间隔时将会错过的传输机会、对初级用户造成的干扰、以及次级用户网络的吞吐量等指标。针对每个指标,基于Sigmoid函数设计次级用户的满意度函数。综合考虑各个指标的满意度,次级用户能够确定最优的频谱感知间隔。 3.设计了基于频谱质量预测的动态频谱接入方案。考虑到传统的随机动态频谱接入方案忽略了接入不同质量的频谱会带来不同的网络性能这一问题,提出一种新型的基于频谱质量预测的动态频谱接入方案,使得次级用户能够优先选择高质量频谱进行接入,从而有效改善网络性能。具体而言： 1)次级用户用非平稳隐马尔科夫模型准确的描述频谱状态的变化过程及自身的频谱感知过程,并通过贝叶斯推断估计非平稳隐马尔科夫模型的参数。 2)通过分析贝叶斯推断获得的非平稳隐马尔科夫模型参数,次级用户对频谱的空闲时间长度、次级用户在频谱上进行频谱感知的检测概率和误警概率进行预测。综合上述指标,定义新型的频谱质量评价指标从而实现次级用户对频谱质量的预测。 3)设计一种新型的基于频谱质量预测的动态频谱接入方案。方案中,次级用户依据频谱质量预测结果对所有频谱进行排序。当需要接入频谱时,次级用户优先选择质量较高的频谱进行频谱感知和接入。
[Abstract]:Spectrum sensing, spectrum decision, spectrum sharing, and spectrum moving are the four major functions of cognitive radio. Secondary users discover spectrum holes through spectrum sensing, and use the three functions of spectrum decision, spectrum sharing and spectrum removal to exploit spectrum holes. Traditional spectrum sensing, spectrum decision, spectrum sharing and spectrum removal In order to solve the above problems, researchers have proposed spectrum prediction technology to predict the spectrum state of the future time by using different methods. In addition, how to predict more spectrum parameters and make full use of these parameters to further improve the overall performance of cognitive radio networks is one of the difficulties to be solved in the field of spectrum prediction technology. This paper aims at the above difficulties to predict the spectrum of spectrum in cognitive radio networks. The main contents and innovations include:
1. the problem of cooperative spectrum state prediction in cognitive radio networks is studied. Recognizing the problem that the local spectrum state prediction method of individual secondary users has limited prediction accuracy, we make full use of the spontaneous cooperation between secondary users, and propose a new cooperative spectrum state prediction method to improve the spectrum state precondition. The accuracy of the test.
1) the cooperation process between secondary users is modeled as a cooperative game process, and a new cooperative group formation algorithm is proposed based on the cooperative game theory to maximize the frequency prediction accuracy of the secondary users. In each group, a secondary user is selected as the head leader to collect the local spectrum of other users in the group. The state prediction results are obtained and the result of cooperative spectrum state prediction is obtained through data fusion.
2) the proposed cooperative spectrum state prediction method is extended to the cognitive radio network composed of any multiple primary users, and the universality of the algorithm is improved. Based on the task allocation idea, the secondary users are classified, and each class of users can only predict the cooperative spectrum state of the authorized spectrum of a primary user. The efficiency of cooperative spectrum state prediction in multiple primary user scenarios is reduced.
2. the secondary user spectrum sensing interval optimization method based on the spectrum state duration prediction is studied. The traditional spectrum sensing technology requires secondary users to perform a spectrum sensing at the beginning of each slot. And the real spectrum data shows that any spectrum state continues to a certain number of slots at a certain probability. At the same time, the optimal spectrum sensing interval of secondary users, that is, how many time slots interval between secondary users' two spectrum sensing, has become a subject worthy of study.
1) secondary users use a hidden Markov model to describe the change of spectrum state, and predict the duration of spectrum state by analyzing the model.
2) based on the prediction results of the duration of the spectrum status, we can further predict the transmission opportunities that the secondary users will miss, the interference to the primary users and the throughput of the secondary user network, and design the satisfaction function of the secondary users based on the Sigmoid function. Considering the satisfaction of each index, secondary users can determine the optimal spectrum sensing interval.
3. the dynamic spectrum access scheme based on the spectrum quality prediction is designed. Considering that the traditional random dynamic spectrum access scheme ignores the different network performance with different quality of the spectrum, a new dynamic spectrum access scheme based on the spectrum quality prediction is proposed, so that the secondary users can be selected first. The high quality spectrum is accessed to effectively improve network performance.
1) the secondary user accurately describes the change process of the spectrum state and its spectrum sensing process using the non-stationary hidden Markov model, and estimates the parameters of the nonstationary hidden Markov model by Bayesian inference.
2) by analyzing the parameters of the non-stationary hidden Markov model obtained by Bayesian inference, the secondary user has the free time length of the spectrum, the secondary user predicts the detection probability and the false alarm probability of the spectrum perception on the spectrum, and defines the new spectrum quality evaluation index to realize the secondary user's spectrum quality. Prediction of quantity.
3) a new dynamic spectrum access scheme based on spectrum quality prediction is designed. In the scheme, the secondary user sort all the spectrum according to the spectrum quality prediction results. When the spectrum is needed, the secondary user selects the high quality spectrum for spectrum sensing and joining.
【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN925

【参考文献】