【作者】 姜园；

【导师】 仇佩亮；

【作者基本信息】 浙江大学 ， 通信与信息系统， 2004， 博士

【摘要】 信息技术在现代军事领域占有越来越重要的地位，成为决定战争胜负的一个关键因素。信息战已经成为现代战争的主要作战形式之一。以控制电磁频谱为目的的电子战是信息战中最重要的核心部分。通信对抗是电子战的重要组成部分。 应用于军事通信对抗的信号处理理论发展非常迅速，这得益于两个方面的动力：其一，军事通信的技术和手段不断更新，在数字化的基础上逐步走向软件化、智能化、宽带化和网络化，出现了自适应跳频、突发通信、宽带调制和复杂编码等新的实用技术，推动了具有很强针对性的信号侦测和处理领域的算法研究；其二，现代信号处理的三大热点——谱估计、高阶统计量方法、时频分析的理论和技术日臻完善，并逐渐应用于通信对抗领域。此外，全数字接收机和软件无线电技术的应用和发展为现代信号处理技术提供了应用平台。 本文研究军事通信对抗中的信号处理问题。以构建对敌方电台进行侦收和指纹识别的系统为目的，本文提出了多种算法，仿真实现了复杂电磁环境中的电台信号提取、信号调制模式识别、扩频信号检测、扩频参数估计，并实现了电台的指纹识别。本文涉及非合作条件下的通信信号处理、非平稳信号分析与处理，并将高阶累积量方法、模式识别、聚类分析和神经网络等多个领域的技术用于信号的分析与处理。 本文首先给出了研究背景、研究目标和系统总体方案，然后根据总体方案的各个模块在各章讨论了相关的算法、仿真实现和性能分析。 接收到多个电台的混合信号后，首先通过信源数目估计、波达方向估计、波束形成和信号聚焦模块，将不同电台的信号分开，并将噪声、电台的暂态信号与通信信号分离。本文提出了改进的基于Gerschgorin理论的信源数目估计算法，提出了基于奇异性的信号聚焦算法，并仿真实现了信号预处理的全过程。 其次，对通信信号，检测是否有扩频信号存在并估计扩频参数；如果是非扩频的一般调制信号，则识别其调制模式。本文提出了一种信号调制模式识别算法，能同时识别7种模拟调制信号(AM、DSB、USB、LSB、CW、FM、AM-FM信号)和7种数字调制信号(2ASK、4ASK、2FSK、4FSK、2PSK、4PSK、16QAM信号)。对直序扩频(DSSS)信号，本文提出了一种假设检验算法检测其存在性；利用DSSS信号的循环平稳特性，提出了估计DSSS信号符号周期的方法。对跳频(FH)信号，本文提出了基于多跳自相关的信号检测方法和基于时频分析的跳频参数估计方法。 再次，利用暂态信号和通信信号的细微特征进行电台的指纹识别。对暂态信号，在没有训练样本的条件下，本文提出了综合聚类算法实现信号的盲分类；在有训练样本的情况下，提出了一种信号特征参数的提取方法，能够更好地表现不同信号类之间的差异，并利用神经网络实现了分类。此外，提出了分析通信信号细微特征的思路，并尝试用高阶统计量方法提取特征参数并识别电台。更多还原

【Abstract】 Information technique plays more and more important role in advanced military fields. It becomes a key factor determining the war victory or defeat. Information warfare is now a main battle form of the modern war. Electronic warfare, whose intention is to control the electromagnetism, is the most important core subsystem of information warfare. Communications Counter-Measures is a main part of electronic warfare.Signal processing theory using in military Communications Counter-Measures has been making great progresses. This is by two reasons: the first, as the development of military communication technologies, it transit from digital to software implementation, intelligence, Broad Band implementation and network implementation. Many new technologies such as adaptive frequency hopping, burst communication, broadband modulation and complex coding accelerate the research about pertinence algorithms of signal sense and signal processing. The second, three hotspots of modern signal processing-spectrum estimation, High-Order Statistics (HOS) and time-frequency analysis theory become more and more consummately and being used in communication jamming and anti-jamming field. Besides, the usage and the development of the all-digital receiver and the software radio technology provide the application platform for modern signal processing technology.This dissertation discussed the signal processing theory and technology using in military Communications Counter-Measures. In order to construct a system to sense and recognize the enemy transceivers, we proposed many algorithms to emulate signal extraction, signal modulation recognition, blind spread spectrum signal detection and parameter estimation, and the fingerprint identification of the transceiver. It deals with the communication signal processing and non-stationary signal processing under non-cooperation situations. Many technologies such as HOS, pattern recognition, clustering and Neural Network (NN) are used here for signal processing.Research background, research goal and the whole system scheme are given in the first chapter, then all kinds of algorithms, emulations and performance analyses of the subsystems are given in chapters later.After we received the mixed signals of transceivers more than one, we estimate the number of signals and the directions of arrival (DOA), form the beams of one transceiver, focus the partition points between the noises, the transients generated when a transceiver turned on and the communication signals. Here we proposed two algorithms, one is a improved algorithm which can be used to estimate the number of signals based on the Gerschgorin theory, the other is a signal focusing algorithm based on the singularities of signals. The whole preprocessing procedure is emulated here.For communication signals, we detect if there exist spread spectrum signals such as Direct Sequence Spread Spectrum (DSSS) or Frequency-Hopped (FH) signals and estimate their parameters if the DSSS or FH signals exist. If there exist normal communication signals, we identify the modulation types. Here a modulation recognition algorithm is proposed which can be used to identify 7 kinds of analog modulations such as AM, DSB, USB, LSB, CW, FM, AM-FM and 7 kinds of digital modulations such as 2ASK, 4ASK, 2FSK, 4FSK, 2PSK, 4PSK, 16QAM. For DSSS signals, we proposed two algorithms, one is a hypothesis test algorithm that can be used to detect if there are DSSS signals, the other is a symbol period estimation algorithm using the cycle-stationary feature of DSSS signals. For FH signals, we proposed two algorithms too, one is a signal detection algorithm based multiple-hop autocorrelation (MHAC) of FH signals, and the other is a parameter estimation algorithm based on time-frequency analysis.The transients generated just when a transceiver turned on and the fine features of communication signals are regarded as the fingerprints of a transceiver. For transients, we proposed two algorithms, one is a improved clustering algorithm which can be used to classif更多还原