通信设备射频指纹产生机理及其个体识别技术研究

Communication device recognition via Radio Frequency Fingerprinting (RFF) extracted from communication signals is a physical layer approach for communication system security. Different communication device exhibits different Radio Frequency Fingerprinting which can be used for identification and authentication. With the development of Cognitive Radio Network, communication device that has the ability of cognition and reconstitution has been widely used. Nowadays, there is no research about the Radio Frequency Fingerprinting for cognitive communication device. We don't know whether there is Radio Frequency Fingerprinting in cognitive communication device, and how to extract the exact and stable Radio Frequency Fingerprinting. Therefore, we want to give the answer. Firstly, according to the structure of Software Defined Digital Radio, the common mathematical model of radio signal and abstract mathematical model of Radio Frequency Fingerprinting will be established. According to the mathematical model, the mechanism of Radio Frequency Fingerprinting will be discussed, and then the influence factor, such as modulation, hardware circuit and their tolerance will be revealed profoundly. Secondly, the evaluation criterion and mathematical model for the characteristic of Radio Frequency Fingerprinting will be established. According to the mathematical model, the characteristic of Radio Frequency Fingerprinting will be analyzed in different transform domain, and then by using the technology of multiscale signal analysis, the effective and difference feature set of Radio Frequency Fingerprinting will be extracted and selected. Finally, by establishing the data acquisition platform and standard database, a high efficiency and low error rate individual recognition model for communication device will be proposed, which is based on deep learning network. The research result will promote the development of physical-layer security theory and technology of Cognitive Radio Network. Therefore, it has theoretical value and practical significance.

通过分析射频信号来提取通信设备的射频指纹进行设备识别是一种保护通信系统安全的物理层方法，可以用于无线设备的身份识别和接入认证。随着认知无线网络的不断发展，大量具有认知和重构能力的通信设备被广泛应用，此类设备中是否存在以及如何提取识别精确度高、稳定性好的射频指纹，目前尚无相关研究成果。本项目首先从软件无线电数字发射机结构入手，构建射频信号产生的通用数学模型和射频指纹的抽象数学模型，分析射频指纹的产生机理，深度揭示调制方式、硬件电路及其“容差”等因素对射频指纹的影响。其次，构建射频指纹特性的评价标准和数学模型，在多变换域上分析射频指纹的特性，通过多尺度精细信号分析技术，提取和选择载荷通信设备差异的射频指纹集。最后，搭建科学数据采集平台，建立标准数据集，构建基于深度学习网络的高效率、低错误率通信设备个体识别模型。本项目对推动认知无线网络物理层安全理论与技术的发展，具有重要的理论价值和实际意义。

基于物理层的射频指纹方法利用发射机射频信号的细微差异来区分不同个体，具有难以克隆、伪造的优点，有着广阔的应用前景。本项目首先从软件无线电数字发射机结构入手，构建射频指纹数学模型，分析发射机射频组件和无线信号信道传播环境对射频指纹的影响，深刻揭示射频指纹产生机理，分析其影响因素。其次，通过研究射频信号在小波、分数阶傅里叶变换，以及多重分形等不同变换域上的特性，提取和选择具备通信设备差异的射频指纹集。最后，利用软件无线电平台构建民航飞机、无人机、WiFI设备等射频信号数据集，结合基于深度学习网络的识别模型，实现通信设备的高效、精准识别。本项目对推动认知无线网络物理层安全理论与技术的新理论和新方法是有价值，有意义的。