实际复杂模型失配场景下的稳健信号检测技术研究

With the fast development of information technology, there are more and more electromagnetic signals. So the real electromagnetic environment gets more and more complex, which brings the issue of practical scene and model mismatch and drastical degradation of detectors. In order to solve the problem, this project extends the research based on the large scale practical experiments. The researches contain the areas below：1) Robust Blind Detection. We make full use of underlying information of observations in transform domain and avoid the relying on existing theory model. We design a method based on observations which has the inherent advantages of invulnerability to noise uncertainty and model error. 2) Large Scale Detection. We make use of correlation between different tests and mine the observations deeply. Based on these, we design the optimal method overall. We further improve the method to make it works well in Large Scale scene. 3) Cooperative Large Scale Detection in Varying and Imperfect Scene. We consider the confidence of the detection results of each node, and further introduce nonlinear quantization and multilevel coding into cooperative detection to design an optimal scheme which has the advantages of both validity and reliability. The research of this subject is expected to provide technical support for the application of signal detection in practical complex scenes.

随着信息技术的发展，空间中的电磁信号日益增加，实际电磁环境越来越复杂，导致现有理论模型与实际场景失配，检测性能急剧下降。针对此问题，课题以大规模线路实测数据为基础，从以下方面展开研究：1）稳健盲检测。充分利用采样数据在变换域中的潜在信息，设计数据驱动的检测方法，摆脱对理论模型的依赖，能有效对抗噪声不确定度和模型误差等高动态环境带来的影响，具有良好的稳健性和普适性。2）Large Scale检测。深入挖掘海量观测数据，解决现有方法中各检测之间相关信息缺失的问题，设计全局最优的检测方法，并进一步优化和简化，使之能适应Large Scale场景。3）动态非理想环境下的协作Large Scale检测。充分考虑各节点检测结果的可信度，同时引入非线性量化和Multilevel编码技术，设计兼顾有效性和可靠性的协作机制，实现联合最优的协作检测。课题的研究有望对信号检测在实际复杂场景中的应用提供技术支撑。

随着信息技术的发展，空间中的电磁信号日益增加，实际电磁环境越来越复杂，导致现有理论模型与实际场景失配，检测性能急剧下降。针对此问题，课题从以下方面展开研究：.1、稳健盲检测。充分利用采样数据在变换域中的潜在信息，设计了FGOF方法、频域卡方检测算法，摆脱对理论模型的依赖，能有效对抗噪声不确定度和模型误差等高动态环境带来的影响，具有良好的稳健性和普适性。.2、Large Scale检测。设计了多目标检测算法，充分利用各检测子带之间信息的相关性，能有效检测出多个目标信号，并进行了相应的性能分析。.3、动态非理想环境下的协作Large Scale检测。针对实际复杂环境，包括各检测节点噪声类型不同，噪声强度不同，衰落不同设计了协作方法，进行协作检测，能有效提升检测性能，并具有稳健性高的特点，进一步进行了详细的性能分析。.课题的研究成果在实际场景中有潜在的应用价值。