【作者】 梁辉；

【导师】 赵晓晖；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2011， 博士

【摘要】 资源分配问题是确保通信系统正常运行且有效利用系统资源的关键技术之一。然而,在面对近些年出现的为解决频谱资源日益紧张与现实频谱利用率过低之间矛盾的认知无线电这一全新技术时,由于认知系统的动态多变特性,使传统原有的资源分配算法难以满足这一新技术的需要。如何在认知系统中合理与高效地分配系统资源已成为当前无线通信领域里研究的热点课题,同时该项技术也必将为无线通信技术的下一步发展提供新的机遇与挑战。当前,针对认知无线电系统这一全新领域的资源分配算法主要包括有基于经典凸优化、博弈论、图着色、协作方式及智能优化等理论的相关算法。这些算法虽然在一定程度上满足了认知系统在某一时刻的优化目标,但从算法对认知系统动态特性的适应性上来看,却并非是最佳的系统描述与求解方法。为适应认知无线电这一新技术的相关特性,本论文采用了多种准则从不同角度对该问题进行讨论、分析和研究,并提出了相应的系统模型与解决方案。论文的前两章阐述了本文的研究意义和背景,介绍了目前国内外对认知无线电资源分配问题的研究现状,并详细分析了现存算法的一些主要不足。简要介绍了认知无线电基本概念与OFDM的基本原理以及采用OFDM技术作为认知无线电系统实现的技术手段的诸多优点,给出认知系统下资源分配问题所具有的特点。为适应认知系统网络环境多变的特性,同时兼顾系统中各用户间的公平性,本论文在第3章提出了基于公平度门限的认知无线电系统资源分配算法。由于传统通信系统中多要求各用户间具有严格的比例公平性,这在一方面降低了系统的整体容量,另一方面又因运算复杂而减少了有效传输时间。相对于传统通信系统,认知系统具有伺机通信的特点,即当主用户暂时未使用该频段时,认知系统可以暂时接入系统进行数据传输,而当主用户重新占用该频段时,其必须归还原有频段的使用权。这就容易造成认知系统资源分配策略计算还未完成,主用户就已经回到原来频段的现象。过于复杂的资源分配算法占用了认知用户宝贵的传输时间,使认知系统的性能大大下降。因此,对于认知系统来说,能够获得较多的传输机会比用户间精确的公平要重要得多。在权衡了用户间公平性与算法复杂度之间的关系后,本文提出了公平度门限的概念,算法在牺牲了部分公平性的同时,获得了更高的系统容量,且在系统功率分配阶段采用了粒子群优化方法,大大加快了算法的收敛速度,使之更加适应认知系统的动态特点。仿真结果表明,本章所提算法在保证用户间一定公平性的前提下,有效地提高了系统的整体性能,且算法具有较快的收敛速度。论文第4章根据认知系统受主用户活动强度影响显著的特点,提出了一种基于主用户活动行为,同时综合考虑主用户中断概率约束的资源分配算法。在认知系统中,认知用户必须实时监测主用户对频谱的占用情况,从而做出相应的资源分配策略。然而,当主用户活动较为频繁时,认知用户所做出的反应往往跟不上主用户的这种快速变化,因此需要从一种新的角度来刻画主用户的活动情况,从而做出更加适合认知系统特点的资源分配算法。在本章所提算法中,将主用户在各频段上的活动情况通过相应的活动概率来加以定义,然后从统计学的角度给出认知用户在各频段上由于主用户活动所造成的速率损失,从而得到该频段认知用户实际能够获得的速率。同时,为确保主用户的传输质量,论文中引入了主用户数据传输时的中断概率作为该优化问题的限制条件进行求解。该算法在综合考虑了主用户的活动强度对认知系统的影响的同时,又通过引入信息中断概率来保证主用户的QoS要求。通过计算机仿真实验结果,可以看出该算法无论从系统容量和对网络的适应性上都较之前文献中提出的算法有了较大的提升。在网络环境快速多变的认知无线电系统中,稳定与可靠的传输显得更为重要。论文在第5章中通过引入经济学中的组合投资理论,将认知环境下的资源分配问题类比为证券市场中的最佳投资选择问题,从而获得了在方差最小化意义下稳定的数据传输方案,同时以系统用户间干扰作为整个优化问题的约束条件,以限制认知系统对主用户的有害干扰。该所提出的算法中,将待分配的功率看作证券市场中的“总资产”,各子载波上的信道容量看作功率分配的“收益”。因此该问题的优化目标即在给定系统期望速率的前提下,使系统偏离期望速率的波动达到最小。这就与证券投资市场中的“风险最小化”问题从其数学本质上来看是同一个问题。因此,我们利用经济学中的组合投资理论,将其应用到认知系统的资源分配问题中来,同时引入用户间干扰门限作为约束条件以保证主用户的通信质量。仿真结果表明,该算法在保证系统稳定传输的前提下,较文献中前人提出的算法在主用户保护上更加有效,更符合认知系统自身的特点。为使认知系统在整个传输过程中获得均值意义下的最大传输速率,论文在第6章中利用动态规划的思想对认知系统的资源分配问题进行了分析与研究,并通过严格的数学推导得出最终的动态规划框架下的迭代表达式。从认知系统以时隙作为单元进行数据通信的角度来看,它刚好符合动态规划理论中在各阶段做出决策的特点,且认知系统动态变化的特点也可通过动态规划中状态间的转移关系较好的进行刻画,故可以采用动态规划的方法对认知系统的资源分配问题进行求解。该算法首先通过离散时间马尔科夫链模型对主用户在各子载波上的占用情况进行描述,然后给出各种网络状态之间的相互转移关系,同时为保证主用户的传输质量,引入主用户速率损失模型作为约束,最后给出了具体的迭代数学表达式用以算法的最终实现。仿真结果表明,本章所提算法在使认知系统整个传输过程中的平均速率达到最大化的同时,有效地保证了主用户的传输速率要求,为认知环境下的资源分配问题提供了一种新的方便、快速与实用化的解决方案。最后,第7章对论文进行了总结,并展望了下一步的研究工作。更多还原

【Abstract】 Resource allocation is one of the key technologies to guarantee communication system to work normally as well as utilize the system resource efficiently. However, an entirely new technique, i.e. cognitive radio which is supposed to resolve the contradiction between the scarcity of wireless frequency spectrum and low utilization of the existed spectrum, has hardly satisfied the needs of this technique due to its fast dynamic feature with former resource allocation algorithms. How to reasonably and efficiently allocate resource in the cognitive radio system is becoming a new hot research spot in the research area of wireless communication. At the same time, this technique also can surely provide many opportunities and challenges for the future development of communication technologies.Recently, the existing cognitive radio resource allocation algorithms are mainly based on classical convex optimization theory, game theory, graph coloring theory, cooperative manner, as well as intelligent optimization theory etc. Although these algorithms can partially satisfy some optimal objectives in one timeslot, they are not the most suitable methods to describe system physical feature to this new problem from the point of view of cognitive radio dynamic feature. In order to fit the new characters of this technology, this paper tries to use multi-norms to discuss, analyze and study this problem from different profiles, and gives a corresponding system model and some solutions.The research significance and background, the current status on cognitive radio resource allocation research both at home and aboard are presented in the first two chapters. And the analysis of the main weakness of the existing algorithms and basic concept of cognitive radio under OFDM principle are discussed briefly. Then the merits of implementing cognitive radio system by OFDM technology are expounded. At last, the features of cognitive radio resource allocation problem are given.In order to adapt the fast changing feature of network environment and consider the fairness among users in cognitive radio system, chapter 3 proposes a fairness threshold based cognitive radio resource allocation algorithm. On the one hand, traditional communication system often demands a strict proportional fairness that may cause the whole system capacity drop. And on the other hand, the complexity of traditional algorithms may also lead to the reduction of system available transmission time. Comparing to the traditional wireless communications, cognitive radio has an opportunistic transmission feature. This makes that the cognitive user can use the licensed spectrum bands during the time when primary user does not use it. However cognitive user must release the bands immediately when primary user reoccupies them. It may easily bring the problem that the resource allocation algorithm has not been calculated and the primary users have already come back to the bands. Moreover, more complex algorithms take more precious transmission time, so that the system performance heavily degrades. Therefore, it is clear that much more transmission opportunities are more important than the extreme strict fairness among different users in cognitive radio system. By considering the trade-off between system fairness and algorithm complexity, we propose a fairness threshold concept that obtains more capacity by sacrificing partial fairness in the system. At power allocation stage, we introduce the particle optimization method to resolve this problem. It can largely promote algorithm convergence speed. Simulation results demonstrate that the proposed algorithm efficiently improves whole system performance under a certain fairness condition, meanwhile obtain a fast convergence speed.Due to cognitive radio system is often significantly affected by primary user’s activity, we propose a primary user activity based and primary user transmission outage probability constrained resource allocation algorithm in chapter 4. In a cognitive radio networks, secondary user must watch the spectrum situation timely in order to find out the optimal resource allocation strategy. However, when primary user’s activity is too frequent, secondary user often can not catch up this rapid change. In consideration of this problem, in order to obtain a more appropriate resource allocation algorithm for cognitive radio networks, a new description model for primary user’s activity is needed. Here, we construct a model for primary user’s activity on each frequency spectrum by corresponding activity probability. To obtain the real data rate of secondary user, the data rate loss of each spectrum from a statistics view is calculated. Meanwhile, an outage probability concept as one constraint of this optimization problem is introduced in order to guarantee primary user’s transmission quality. In the computer simulation results, it can be seen that the proposed algorithm efficiently improves the performance both in system capacity and network adaptation compare to the former method.In a fast changing cognitive network environment, it is considered that the stable and reliable transmission is more important. In chapter 5, we introduce the portfolio selection theory in economy to formulize cognitive radio resource allocation problem into a best investment selection problem. And the stable transmission schemes in a minimum-variance sense are given. At the same time, to limit harm interference from the cognitive user, mutual interference concept is introduced as a constraint of this problem. In this algorithm we look the system total power and channel capacity on each subcarrier as total security assets and return of each security respectively. Therefore, our optimal objective is to minimize the variance of system data rate under a given expected transmission rate. From the mathematics, it is the same problem with“risk variance minimization problem”in security market. Hence, we can use portfolio selection theory to deal with resource allocation problem in a cognitive radio system. Meanwhile, communication quality for primary user by mutual interference threshold is efficiently protected. Simulation results show that the propose algorithm can provide more efficient protection to primary user compare to the former scheme at the same stable transmission rate. This improvement makes the proposed algorithm more adaptable for cognitive radio system.In order to maximize a long-term average rate during the whole transmission process in cognitive radio system, a dynamic programming method to analyze the resource allocation problem is proposed, and a mathematic iterative expression is derived in chapter 6. Since in the cognitive radio transmission the data is transmitted in each time slot, it is just correspond with the character of dynamic programming that the decisions are made in each stage. At the same time, the dynamic changing feature of cognitive radio can also be well described by transitional relationship among different states in a dynamic programming framework. Therefore, it is a valuable attempt to deeply discuss the cognitive radio resource allocation problem with dynamic programming method. This method can model primary user’s occupation on each subcarrier by discrete-time Markov chain. And it gives the transition probabilities among different system states. Meanwhile, a primary rate loss model to guarantee primary user’s transmission quality is introduced. Finally, a specific dynamic programming iterative expression is given. In the given computer simulations, it shows that the proposed algorithm not only maximizes the long-term average rate in the whole transmission process but also efficiently guarantee primary user’s request data rate. In addition, it also provides a convenient, fast and practical method for cognitive radio resource allocation.At last, chapter 7 concludes the whole paper and forecasts the next step research work.更多还原