基于自由概率理论的宽带频谱感知及其非渐近分析

项目名称： 基于自由概率理论的宽带频谱感知及其非渐近分析

项目编号： No.61201164

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 电子学与信息系统

项目作者： 王磊

作者单位： 南京邮电大学

项目金额： 25万元

中文摘要： 认知无线电技术是解决当前频谱资源匮乏这一瓶颈问题的有效手段，频谱感知是认知无线电的核心和基础，已引起国内外众多研究机构的高度关注。实际中认知无线电系统要求进行宽带或全频带感知，但目前大部分算法仅适合窄带感知，现有的宽带感知算法性能还无法满足实际需要。因此，跳出现有理论框架的束缚，实现高性能宽带频谱感知，是认知无线电技术走向实际应用的关键。本课题的主要目标是基于自由概率理论（Free Probability Theory）实现高性能宽带频谱感知：其一，利用自由概率理论在非交换空间中实现经典数学理论无法做到的谱分离，设计高性能宽带频谱感知算法；其二，探索执行自由解卷积（Free Deconvolution）算子的快速算法，降低感知复杂度；其三，进行非渐近理论分析，验证新方案在实际有限维样本空间中的快速收敛性。本研究可为认知无线电的实际应用提供理论和技术支持。

中文关键词： 认知无线电频谱感知；随机矩阵理论；自由概率理论；判决融合；分簇选择

英文摘要： Cognitive radio technology is an effective way to solve the current key problem of lacking spectrum resources. Spectrum sensing, which is the foundation of cognitive radio technology, has been widely studied by many research institutions. Practical cognitive radio systems require the wideband or entire-band spectrum sensing. Unfortunately, the majority of current algorithms mainly focuses on the narrow-band spetrum sensing, and hence current wideband sensing algorithms can not meet the practical requirement. Therefore, to realize the high performance of wideband spectrum sensing is an essential step for cognitive radio technology in the future application. Based on the free probability theory, the main goal of this project is to realize the high performance wideband spectrum sensing. Firstly, using the free probability theory which can achieve spectrum seperating in the non-commutative space while the classic math theory can not, we will design an algorithm for high performance wideband spectrum sensing. Secondly, we will seek for a fast algorithm for free deconvolution operator to reduce the complexity of spectrum sensing. Thirdly, we will verify the feature of the fast convergence of this new sensing scheme in a practical limited dimension sample space based on non-asymptotic analysis theory. This research can

英文关键词： cognitive radio spectrum sensing；random matrix theory；free probability theory；decision fusion；clustering and selection