融合认知强化压缩感知与深度学习的扩频通信信号拦截研究

As an important part of electronic countermeasures, the interception of communication signals has become a key factor to determining the outcome of a war. Spread spectrum (SS) is the mainstream of modern military wireless communication technology, and its communication bandwidths and code rates are increasing day by day. This puts a burden on the signal interception based on traditional Nyquist sampling. The adoption of compressive sensing can alleviate this problem theoretically, but for the ever-changing battlefield channels and signals, existing research still lacks in the algorithm capabilities of adaptivity and online processing. In order to break through the above bottlenecks, this project intends to do research on the algorithms and key scientific problems of SS signal interception, with the integration of the ideas of knowledge enhancement compressive sensing and deep learning. On the one hand, the idea of knowledge enhancement compressive sensing can reduce the signal sampling rate and processing complexity while adding cognitive adaptability to the SS signal interception. On the other hand, the deep learning idea can enhance the signal feature extraction and online processing capabilities for the algorithm. Based on the above two points, in lack of signal prior knowledge, the algorithm results of this project will be able to detect the SS communication signals, estimate the signal parameters and intercept the symbol contents online, adaptively and accurately, at low frequency sampling rates. The research of this project will contribute to the research and development of military electronics in our country, so the project has a great research value.

作为电子对抗的重要组成部分，通信信号的拦截已成为决定战争胜败的关键。扩频是现代军用无线通信技术的主流，其通信带宽和码率与日俱增。这为基于传统奈奎斯特采样的信号拦截带来负担。压缩感知的使用在理论上缓解了这一问题，但针对多变的战场信道与信号，现有研究在算法的自适应与在线处理能力方面仍存在不足。为突破上述瓶颈，本项目拟融合认知强化压缩感知与深度学习思想，研究扩频通信信号自适应拦截的算法与关键科学问题。一方面，认知强化压缩感知的思想能够在降低信号采样率与处理复杂度的同时，赋予扩频信号拦截认知自适应性的特点；另一方面，深度学习思想能够增强算法的信号特征提取能力和在线处理能力。基于以上两点，本项目的算法成果将能够在缺乏信号先验知识的条件下，以低频的采样率，在线、自适应且准确地检测扩频通信信号、估计信号参数并截获码元内容。本项目的研究将为我国军事电子领域的研究和发展做出贡献，因此项目具有重要的研究价值。

作为电子对抗的重要组成部分，通信信号的拦截已成为决定战争胜败的关键。扩频是现代军用无线通信技术的主流，其通信带宽和码率与日俱增。这为基于传统奈奎斯特采样的信号拦截带来负担。为了突破传统算法的局限，是我国在未来的战场上处于信息对抗的领先地位，本项目以跳频信号与直接序列扩频信号为例，融合认知强化压缩感知与深度学习理论，研究了扩频通信信号拦截原理与算法等关键科学问题。研究结果包括扩频通信信号的检测、参数估计与码元检测算法。其中信号的检测算法能够在采样率小于奈奎斯特采样率数倍的情况下，仍能够取得良好的检测率，且漏检率相对于传统的随机压缩检测算法降低1个数量级左右。信号的参数估计算法能够以压缩采样率，正确地估计跳频信号与直接序列扩频信号的时频信息与相关参数。即使在低信噪比的情况下，所研究的算法对相关信息估计的准确度也有着明显优势。信号的码元检测算法能够以较高的压缩比自适应采样并检测信号码元，尤其针对二进制高斯滤波频移键控跳频信号，当压缩感知的压缩比为5时，码元检测的准确率能够接近非压缩采样。从科学意义角度分析，本项目解决了解决传统扩频信号检测算法大多需要高采样率，而现有基于压缩感知的算法自适应能力较差的问题，现有压缩感知信号参数估计算法受信号模型影响大且算法复杂度高，以及现有基于深度学习的扩频信号参数估计算法采样率高，而传统扩频码元检测算法采样率高，现有基于压缩感知的算法大多需要对信号进行重构，且认知自适应能力不足的问题。本项目共发表高水平论文9篇，其中SCI检索5篇，EI检索8篇，中文核心期刊1篇；参加国际会议3次；申请发明专利4项；与所在学院其他教师共同培养博士研究生1名；培养硕士研究生3名。