【作者】 刘丽霞；

【导师】 彭宇行；

【作者基本信息】 国防科学技术大学 ， 计算机科学与技术， 2012， 博士

【摘要】 频谱接入是无线认知网络研究的核心问题，通过认知用户对授权频谱的接入共享来实现频谱利用率的提高，从而满足越来越多无线应用的频谱资源需求，缓解频谱资源缺乏和已授权频谱利用率低下的矛盾。本文围绕频谱接入的高效资源利用，分别以频谱传输有效性、信道成功传输概率、频谱空间复用度和用户需求匹配度等目标展开研究，主要工作和创新成果包括：针对无线认知网络数据量不大、节点存储和运算能力有限、频谱资源可用性快速变化的特点，本文分析无线认知网络中有效信息查询问题后，提出了快速查询动态数据的极大点分区查询算法MP-DRQ (Maximum Point-Different RegionQuery)，达到了提高数据传输有效性和频谱利用率的目的。该算法的主要思想是：提出“当前极大点”概念，以当前极大点为依据将信道空间划分为控制区域、被控区域和自由区域，对动态信道数据进行分区域查询处理。实验表明，与BNL和D&C算法相比，MP-DRQ可以大量减少查询运算量。针对无先验知识条件下的频谱分配问题，本文在分析历史信息对频谱分配所产生影响的基础上，提出了基于信息素的频谱分配算法PSA (Pheromone basedSpectrum Allocation)，达到了提高成功传输概率、优化频谱使用的目标。该方法借鉴蚁群优化理论中的信息素概念，将频谱分配中携带具有时效特性的成功传输概率的广播信息看作“信息素”，通过“信息素”的更新不断调整频谱分配方案，持续这种渐进迭代过程，直到收敛到一个满足网络总的成功传输概率需求的频谱分配方案。实验表明，该算法比其他常用算法，如随机选择策略、固定选择策略和贪婪选择策略，在提高成功传输概率方面体现了较大的性能优势。针对如何让更多用户共享转发交易中频谱资源的问题，本文在分析转发交易模式下频谱奖励共享问题的基础上，提出了转发交易模式下频谱共享策略SSS-RT(Spectrum Sharing Strategy for Relay Trading)。该算法以最大化频谱空间复用度为目标，以设置满足最小传输速率的发射功率为技术途径，组建合作共享群，定义贡献度衡量各成员的转发贡献和共享资格。此外求解能满足最小数据传输速率的功率上下界，并借鉴迭代注水思想，采用多轮功率设置和调整，使得满足最小传输速率的用户数最多。实验表明，采用SSS-RT的转发交易模式可以大幅提高频谱空间复用度。针对多重服务属性约束下的频谱交易问题，本文在分析频谱多重属性对频谱交易影响的基础上，以优先级约束下最大二分匹配为目标，提出了面向异构服务的频谱交易算法HSO-ST(Heterogeneous Service Oriented Spectrum Trading)。首先对服务员的频谱服务属性进行具体定义，再根据频谱供应集合建立频谱交易的服务空间；然后基于频谱需求与服务空间以及频谱供应与服务空间的对应关系，按照属性匹配度分级匹配，建立供需关系对应的二分图，并根据HSO-ST算法实现最大二分匹配。实验表明， HSO-ST算法比其他频谱交易方案提高了认知用户的需求匹配率。更多还原

【Abstract】 Spectrum access is the core part of wireless cognitive network research, the mainidea of which is that the cognitive users share the spectrum of the lisenced users toimprove the spectrum utilization ratio, satisfy the rapid-developing spectrum demandand ease the contradiction of spectrum-scarcity and lower-utilization of the lisencedspectrum.In this paper we focus on the problem of efficient spectrum access and carryout research with the targets of transmission availability, channel successfultransmission ratio, degree of spectrum share and demand matching ratio respectively.The main contributions include:Considering the features of wireless cognitive networks, such as small-scale traffic,limited storage and computing ability and rapid-changing spectrum availability, thispaper analyzes the problem of useful information query for wireless cognitive networks.We propose the Maximum Point-Different Region Query algorithm (MP-DRQ) whichcan query the dynamic data quickly. It achieves the research target of improvingtransmission availability and spectrum utilization. The concept of current maximumpoint is proposed。The channel space is divided into dominate region, dominated regionand free region according to current maximum point(s). It fits for the dynamic dataquery problem. Simulation results show that the query time with MP-DRQ can bereduced comparing with those of BNL and D&C.Considering the spectrum allocation problem without any prior knowledge, wepropose the Pheromone based Spectrum Allocation algorithm (PSA), which is based onthe analysis about the effect of history information on spectrum allocation. It achievesthe research targets of improving successful transmission ratio and optimizing spectrumusage. PSA gets inspiration from the concept of pheromone in the ant colonyoptimization theory. The timely broadcast information with successful transmissionratio can be regarded as pheromone during the spectrum allocation procedure. Adjustthe spectrum allocation strategy with the update of pheromone. This process will notstop until the algorithm converges to a spectrum allocation strategy which can satisfythe demand. The simulations show that PSA has the obvious performance advantage ofimproving the successful transmission ratio, comparing with random spectrumallocation, fixed spectrum allocation and greedy spectrum allocation.Considering how to make more users to share the spectrum in the relay-tradingmode, we analyze the spectrum sharing problem in the relay-trading mode. Based on itwe propose the spectrum sharing strategy for relay trading. The target of this algorithmis to maximize the degree of spectrum share and the method is to set the power whichsatisfies the minimum transmission rate. We construct a cooperative sharing group anddefine the contribution metric to measure the relay contribution and sharing qualification of each group. We find the power bounds which can satisfy the minimumtransmission rate. Learning from the water filling algorithm, we set and adjust thepower for many cycles, which can make more users satisfied. Simulation results showthat the relay trading mode with SSS-RT can improve the degree of spectrum sharingobviously.Considering the spectrum trading problem with multi-service-attribute, we proposethe Heterogeneous Service Oriented Spectrum Trading (HSO-ST) algorithm based onthe analysis of effect of multi-service-attribute on spectrum trading, the target of whichis to maximize the matching ratio of secondary users with the priority restriction. Firstlywe give a detailed definition of spectrum service character and construct the servicespace according to spectrum supply set. Based on the relationships of spectrum demand,spectrum supply and service space, we build the bipartite matching graph according toattribute matching degree, and obtain the maximum bipartite matching with HSO-ST.Simulation results show that the matching ratio with HSO-ST can be improved by atleast10%comparing with other spectrum trading strategies.更多还原