Time-Frequency Analysis Based Detection And Parameter Estimation For Hybrid Spread Spectrum Signals

Communication countermeasure is one of the most important parts in electronic countermeasures, which is the warfare in radio area. Since spread spectrum (SS) techniques are widely utilized in communication countermeasure systems, researches on effective detection and parameter estimation methods of SS signals have great practical significance and remarkable application value. This dissertation mainly deals with the blind detection and parameter estimation of hybrid SS signals under the framework of time-frequency analysis (TFA). The main achievements are as follows:1. The characteristic property of white Gaussian noise is derived in S transformation domain. Based on the analysis results, a linear TFA based detector is proposed for the detection of the hybrid SS signals. The detector can provide time-varying information in addition to detecting whether the signal is present or absent in low SNR conditions, which is helpful for the parameter estimation task.2. The performace of present parameter estimation methods for direct sequence SS signals are analyzed in the time-invarying parameter estimation of the hybrid SS signals. Then a reference signal generation method is proposed, in which a nonlinear process is used to produce the reference signal which has the same hop parameter as the original signal and its frequencies could be estimated by the cyclic spectrum. The analysis indicates that the noise is suppressed in the operator and this is helpful for the using of TFA methods.3. Combining the idea of the channelized receiver and the auditory masking phenomena, a new hop rate estimation schemes for the hybrid SS signals is addressed by employing a novel channelized spectral enhancement preprocessor and TFA. This externally-linear-internally-nonlinear (ELIN) channelized spectral enhancement system can realize interference-suppression and enhance noisy signal without losing detail informations. Based on the reliable signal enhancement results, TFA is used to estimate the hop rate of the hybrid SS spread spectrum signal.4. Based on the analysis of shortcomings of the adaptive TFA methods in the scope of applications, a new parameter estimation method for hybrid SS signals which employs adaptive decimation based Short-Time Fourier Transform (STFT) is proposed. It uses the decimation to realize the time-frequency representation with adaptive window width based on the filter-bank interpretation of STFT. Since the computational burden caused by the direct change of window width is alleviated, the proposed method is easy to implement. Then the method is introduced into Short-Time Hartley Transform (STHT) based on the relation between STFT and STHT. STHT involves only real operations and its transform kernel is the same for both the forward and inverse transforms, which means that any hardware built to compute the forward transform can be used to compute the inverse transform without modification. The two methods are illustrated under the MMSE criteria. The optimal decimations can track the change of the instantaneous frequency (IF) and the minimum factors indicate the hop time parameter.5. Several generalized S transforms (ST) are examined for their estimation performace of IF. Based on the results, two methods are proposed for the parameter estimation of hybrid SS signals, they are:1) Original S transform is improved by adding a Gaussian window width parameter to realize the adaptation of its time-frequency resolution. Then the modification information is used to reduce the estimation error in high frequency region caused by noise.2) Original S transform is improved by using an asymmetrical window instead of a Gaussian one. This asymmetrical window has a sharper taper in the forward direction and a compensating slower taper in the backward direction. Thus the resolution of hopping termination times is sacrificed to improve the resolution of hopping initiation time in the resultant time-frequency spectrum. Since the onset and offest of every hopping can provide the same parameter information, the asymmetrical S transform is better at parameter estimation for hybrid SS signals than that which uses the symmetrical windows.