非协作多认知用户环境下的稳健频谱感知与交叉信道增益估计问题研究

Spectrum sharing can greatly improve spectrum utilization efficiency, and it is an important means to solve the contradiction between spectrum supply and demand. However, with the increase of spectrum sharing users, the non-cooperation among cognitive users pose a serious challenge on the spectrum sharing, which will lead to the reduction of spectrum efficiency and even induce spectrum congestion. This project focuses on the robust spectrum sensing and cross-channel gain estimation, aiming to the spectrum sharing among non-cooperative cognitive users. The main contents include: Establish accurate spectrum hole model by analyzing and estimating the statistical parameters of primary and cognitive users; Design robust spectrum sensing methods by combing parametric and nonparametric means through Turbo iteration and modern statistical processing tools; Estimate cross-channel gains based on factor graph theory by mining the features of variation of interference amounts among cognitive users. Our study will facilitate the spectrum sharing technology to their application in practice, and hence, this work is of great importance in both theory and practice.

频谱共享能极大地提高频谱利用率，是解决频谱供需矛盾的重要手段。然而，随着频谱共享的认知无线电用户的增多，认知用户之间的非协作性等问题对频谱共享带来了严重挑战，这会导致频谱利用率降低甚至引起频谱拥塞。本项目拟针对非协作多认知用户共存环境下的频谱共享问题，围绕稳健频谱感知和交叉信道增益估计两个关键问题开展研究。具体内容包括：分析授权用户和认知用户的统计特性，对其相关参数进行估计，进而建立准确的频谱空洞模型；针对非协作多认知用户共存引起的的复杂干扰、信号随机到达和离开及信号累积功率随机性等因素，利用Turbo迭代处理、现代统计信号处理等工具估计其统计分布，融合参数化和非参数化等手段研究稳健频谱感知方法；通过挖掘多认知用户之间干扰量的变化规律，采用因子图理论等工具研究交叉信道增益估计方法。相关研究成果将对高效频谱共享技术的发展起到良好的推动作用，具有重要理论和实际意义。

本课题首先研究非协作授权用户随机到达和离开的特性，利用半马尔可夫模型的信道状态变化进行建模，同时分析了频谱空洞的边沿（即信道状态跳变时刻）的统计分布特性。然后，通过提出一种迭代算法提高了授权信道状态占空比估计性能，使得所建频谱空洞模型更加准确。接着，围绕稳健频谱感知，针对噪声功率不确定性、授权用户随机到达/离开引起性能平层、次级用户所处环境不一致导致协作感知性能下降等问题，提出了一系列稳健的频谱感知方法。最后，通过挖掘授权用户发射机的发射功率与认知用户发射机的发射功率之间的关系，提出了一种主信道增益和交叉信道增益联合估计方法。