【作者】 王浩；

【导师】 曾凡仔；

【作者基本信息】 湖南大学 ， 信息与通信工程， 2015， 硕士

【摘要】 相比单用户频谱感知技术而言,协作频谱感知技术拥有更好的感知性能,它的工作原理是把多个感知用户的感知结果以适当的策略融合在一起得出最终的判决结果。目前大多数的研究都假定在理想环境下进行,在此基础上尽可能找到网络的最优检测概率以期获得最大的吞吐量,而没有考虑协作频谱感知网络中的感知开销、资源浪费的问题。而且大多数的协作频谱感知融合策略把所有的单感知用户的本地感知结果不加区分的融合在一起处理分析,没有考虑其中恶意用户等带来的干扰和误导。针对这两个问题,本文工作如下:提出具有能量采集功能的分布式协作感知网络中的感知用户自私和感知用户部分合作两种不同情况下的协作感知策略。这些策略建立了感知用户参与感知的概率和收益的函数关系,分别应用非合作博弈的混合策略纳什均衡和基于牛顿迭代定理的最优算法获得相应的最优感知概率值,从而得到不同情况下感知用户的最大收益值。实验仿真结果比较了这两种感知策略的性能,得到了感知用户部分合作情况下的感知策略性能更优的结论。提出一个感知用户的信任度参数值动态更新的融合判决算法,该方法利用感知用户传输结果的反馈值和该用户上一次的本地感知结果,通过贝叶斯准则动态更新感知用户当前的信任度参数值。融合决策中心可以根据当前这些该信任度参数值对整个网络中的感知用户分类,并筛选出可靠的本地感知结果进行最终的融合判决,从而能排除协作频谱感知网络中恶意用户和无关用户对最终融合判决结果的干扰,保证判决结果的准确性。实验仿真结果表明,网络中存在恶意用户和无关用户时,相较于传统的融合判决机制,本文的融合判决算法具有更好的性能。更多还原

【Abstract】 Compared with a single user spectrum sensing technology,collaborative spectrum sensing technology has much higher sensing performance.It integrates multiple users’ sensing results in an appropriate way to achieve the final decision.Existing researches are performed under the ideal condition and do their best to find the optimal network detection probability to get peak throughput.However,those utilities are not discussed as a whole,such as sensing cost,resource waste and so on.Moreover,The general collaborative spectrum sensing strategy integrates all single users’ spectrum sensing results indiscriminately to analyze and process,without taking interference and misleading from the malicious users into account.Based on the above-mentioned two questions,the main work of this paper can be demonstrated as follows:Under the distributed collaborative spectrum sensing network with the function of energy harvesting,two collaborative spectrum sensing strategies,which under different scenarios of selfish user and collaborate user respectively,have been proposed.Those strategies construct the relation between the sensing probability and gain function of the sensing users,and apply the Nash equilibrium of non-collaborate game strategy and Newton iteration theorem to achieve the corresponding optimal sensing probability,thus obtain the maximum earning values under different scenarios.The simulation results show the performance comparison of these two different sensing strategies,and draw the conclusion that the performance under the scenario of collaborative user is much higher.Furthermore,a fusion decision algorithm,which is based on the trust value of sensing user can be upgraded,has been proposed.The algorithm use the feedback of the transmitted result of the sensing user and the native sensing result of this sensing user to upgrade the current trust value of the sensing user with Bayes Criteria dynamically.The collaborative decision center can classify all sensing users in the network according to the current trust value of each sensing user,and select the reliable native sensing results to make the final decision.Hence prevent the final decision from being interfered by those malicious or unrelated users in the collaborative spectrum sensing network,and assure the accuracy of the final decision.When several malicious and unrelated users are existed in the network,the simulation results demonstrate that the proposed collaborative decision algorithm in this paper has much higher performance than the traditional decision scheme.更多还原