基于压缩采样的鲁棒宽带频谱感知方法研究

Cognitive radio is a very important technology to improve spectrum utilization, in which spectrum sensing is the key foundation to successfully achieve cognitive wireless communications. In order to meet the needs of future high-speed broadband services and to solve the shortage problem of the carrier frequency resource, this project will conduct some research of theory and technologies for robust wideband spectrum sensing based on compressive sampling, which will provide theoretical and technical assistant for the spectrum management and utilization for future wireless communications...In this project, we will design a new sampling scheme by combing the multi-coset sampling and the co-prime/nested sampling for power spectrum sensing and develop a new robust fast sparse recovery algorithm for power spectrum reconstruction by leveraging the asymptotic theory of the second order statistics and the convex optimization theory. Meanwhile, we will also study the robust wideband spectrum sensing against non-Gaussian noise, and plan to propose a new sparse spectrum reconstruction algorithm for multiple measurement vectors based on the block optimization theory from a perspective of subspace method. In addition, in order to overcome the limited sample truncation effect, we will discuss new parametric power spectrum estimation via modern spectrum estimation with parametric model of stationary signal. Furthermore, we will study and design a new robust wideband spectrum sensing detector by further exploiting the non-circular characteristics of some communication signals. Lastly, based on the above work, some fundamental theory and methods will be achieved for robust wideband spectrum sensing with compressive sampling, which is very significant to boost the development of spectrum sensing for cognitive radio systems.

认知无线电技术是提高频谱利用率的重要技术手段，频谱感知是实现认知无线通信的关键基础。为了适应未来高速宽带业务需求与解决频率资源紧缺问题，本项目将研究基于压缩采样的鲁棒宽带频谱感知基础理论与关键技术，为未来无线通信频谱的管理与使用提供理论依据和技术支持。. 本项目将采用多陪集采样与互素或嵌套采样相结合的方案，设计新型的高效采样方法；结合二阶统计量的渐进分布特性和凸优化理论，设计新型稳健的快速稀疏功率谱恢复算法；结合块优化理论，采用基于子空间的多测量矢量稀疏支撑集恢复算法，设计非高斯噪声下的鲁棒宽带频谱感知方法；针对有限样本截断效应带来的性能损失，结合现代谱估计模型，讨论新型参数化功率谱估计；结合通信信号非圆特性，研究新型宽带频谱检测方法。最后，通过以上内容研究，形成基于压缩采样的鲁棒宽带频谱感知理论与方法，为频谱感知技术的发展奠定基础。

为了进一步发现更多可利用的频谱空洞及适应未来超宽带与高速率移动无线通信系统的发展，宽带频谱感知技术被认为是认知无线电技术的一个重要发展方向，对未来大数据时代信息传输与共享起着举足轻重的作用。但是，目前宽带频谱感知技术要求认知无线电设备的射频前端模拟-数字转换器具有很高的采样速率，需要花费很大的代价才能制造出满足宽带频谱感知要求的ADC，而且大量的信息存储与处理也需要更多的功耗。因此，为了适应未来移动通信的发展，研究具有低采样率要求的鲁棒宽带频谱感知方法具有重要的研究意义。. 本项目研究了多陪集采样、互素采样和嵌套采样方法，设计了一种基于嵌套采样的宽带频谱感知方法，在保证检测概率的情况下，进一步降低了采样率；结合二阶统计量误差的渐进高斯分布特性，设计了非均匀噪声下稳健的多测量矢量支撑集恢复算法；针对非高斯噪声下的参数估计，基于重加权迭代设计了鲁棒稀疏信号参数估计方法；结合现代谱估计模型，讨论了参数化功率谱估计方法；研究了通信信号的非圆特性，设计了基于嵌套压缩采样的非圆信号的宽带频谱检测方法，仿真结果显示，该方法显著提高了非圆信号的检测概率。. 认知无线电技术是提高频谱利用率的重要技术手段，频谱感知是实现认知无线通信的关键技术。为了适应未来高速宽带业务需求与解决频率资源紧缺问题，本项目基于互素采样、嵌套采样等压缩采样方法，研究了基于二阶统计量的鲁棒宽带频谱感知技术与方法，为未来无线通信频谱的管理与使用提供理论依据和技术支持。