Research On Frequency-Hopping Signal Detection In Impulse Noise

Frequency hopping(FH) is an important spread spectrum technique characterized by strong anti-jamming, optimum spectrum efficiency, low probability of interception and detection, and outstanding compatibility. Therefore, FH signal has been applied widely in military and civilian communications. Effective FH signals detection is of great importance to guarantee accurate information transmission, and has become a hot issue recently. In traditional FH signal processing, the Gaussian distribution is used to model the background noise. Gaussian distribution satisfies the central limit theorem, which can be conveniently applied in signal processing and analyzing. However, recent researches show that the interference usually appears with notable impulses in the actual noise or clutter environment such as multiple access, atmospheric noise, sea clutter and so on. This kind of noise can be described accurately by alpha stable distribution, the probability density function of which has more significant peak pulse waveform and thicker tail. In alpha stable noise environment, the performance of conventional processing methods based on Gaussian distribution will degrade drastically. In view of this question, a systematic study on detection methods of the FH signal in alpha stable noise environment is conducted. The major contributions are outlined as follows:1. The time-frequency analysis(TFA) is a powerful tool for FH signal detection, but in alpha stable noise environment, the performance of the conventional processing methods based on commonly used time-frequency analysis can be reduced or even disabled. In view of this question, detection of FH signals based on data credibility weighting, which combines the theory of cloud model with the short-time Fourier transform(STFT) is proposed in this thesis. The concept of the data credibility is introduced based on the cloud model theory to analyze the uncertainty of received signal. On this basis, the weighting process is implemented to the received signal to improve the performance of time-frequency distribution(TFD). Simulation results show that this method can detect FH signals efficiently, and has better performance of FH signals detection than the fractional lower order statistics as well as the Myriad filter based TFA in alpha stable noise environment.2. A detection method of FH signals based on ML estimation for frequency-hopping parameters by Cauchy distribution is proposed in this thesis. The FH signal is decomposed into the two-dimensional envelope versus frequency plane, and then a maximum-likelihood function based on Cauchy distribution is established to extract the frequency parameter directly. For the short-time stationarity of FH signals, the maximum-likelihood function is windowed in order to estimate the specific values and sequence of frequency-hopping, after that FH signals can be detected with hopping timing and the duration extracted. Simulation results show that the proposed method not only can detect all frequency components, but is more robust than the fractional lower order statistics and the Myriad filter based TFA in strong impulse noise environment.