Study On Detection Method Of Weak Leakage Electromagnetic Signal For Digital System

Detection of leakage electromagnetic signal of digital system is an important work in TEMPEST research, which plays an important role in many applications, such as military, security, commerce, fault diagnosis and so on. Nowadays, when leakage electromagnetic protection technology becomes more and more mature, leakage signal is so weak that has presented new challenges for researchers. On the one hand, the weak leakage electromagnetic signal is usually embedded in the strong background noise. As the environment changes, the background noise would have uncertainty, so the detection system must have the strong adaptability for the environment. On the other hand, weak leakage electromagnetic signal is detected by improving the signal-noise ratio. Traditional methods, such as filtering, amplification, are difficult to meet the requirement of detection. The existing methods generally require some prerequisites, such as stochastic resonance and chaotic oscillator. Therefore, to research a new detection method and to improve existing method remain to be solved.To detect weak leakage electromagnetic signal with known frequency, chaotic oscillator detection method was proposed. Weak periodic signal with the same frequency as driving force can change the phase orbit of chaotic system, and result in the transition from chaotic state to great scale periodic state when chaotic oscillator is in the chaotic critical state. Few documents did research on the reliability of weak signal detection based on chaotic oscillator. This paper analyzed Duffing equation's sensitivity to initial conditions, concluded from a lot of stimulation results that different initial conditions led to different transition process, then different immunity to noise, and finally proposed that a proper selection of initial conditions and input time of weak signal embedded in background noise can effectively improve the detection reliability.To detect weak leakage electromagnetic signal embedded into the non-stationary noise, a nonlinear odd function was proposed, Through analyzing the spectral feature of the square pulse signal, it was found that the spectral feature of square pulse signal was not only decided by the pulse width, but also by the pulse amplitude. The nonlinear function could transform the sinusoidal signal into a square pulse signal with the same period. Then, the spectral density of random frequency components was attenuated more quickly than that of the weak leakage electromagnetic signal with the fixed amplitude after transformation, which could improve the SNR and facilitate a better retrieval of the weak leakage signal after the time-frequency procedure. Furthermore, a moving time-frame approach was introduced to detect the initial time of weak leakage signal.To aim at the detection of the small randomly occurring leakage electromagnetic signal with uncertain frequency, amplitude and phase in broad frequency band, a improved clustering algorithm was proposed. Partitional clustering algorithm and hierarchical clustering algorithm were used to cluster the sample information and the to-be-detected information. A comparative analysis of classes between the sample information and the to-be-detected information was conducted. The new classes correspond to the feature values of weak leakage signal were obtaine. In the signal recognition process, the fuzzy-clustering algorithm was used to eliminate the effects of short-time random noise. The method could realize the real-time online detection.The detection system based on the imoroved clustering algorithm was realized and developed, including software and hardware systems. The mixing method was used to transform the signal in broadband into intermediate band with fixed center frequency, which can lower the design requirement of amplifier and sampling instrument. IF signal after time-sharing transformation was amplified and sampled. Experiment results showed that the detection system could realize the real-time online detection of small leakage electromagnetic signal without prerequisites in broad frequency band.To sum up the above arguments, this paper studied data acquisition, signal detection, software and hardware realization of weak electromagnetic leakage signal under various conditions, proposed or improved three detection methods of weak leakage electromagnetic signal, and realized the real-time detection to small leakage electromagnetic signal without prerequisites in broad-frequency band. Simulation and experiment results showed the proposed methods could recognize weak leakage electromagnetic signal in a complicated environment.