Signal Detection Techniques Under Complex Electromagnetic Environment

With the fast development of information technology, there are more and more electromagnetic communications equipments. The communications signal mode and modulation style are more and more complex, which makes the limited spectrum source more and more crowded and the electromagnetic environment more and more complex. The characteristics of the electromagnetic environment are as follows:(1) There are more and more signals, which improves background noise seriously;(2) Various interferer tend more complex and practical systems are drowns in the non-Gaussian background noise;(3) The background noise is high dynamic fluctuation. The complex electromagnetic environment makes the signal detection technology face the higher requirement and more sever challenges:(1) Signal is often drowned in the noise and presents extremely low spectral density characteristics;(2) The classic signal detection methods do not work well in the non-Gaussian noise;(3) Many existing signal detection methods degrade seriously in high dynamic noise;(4) Many signal detection methods are hard to satisfy the real time requirement. This paper carries out research on the challenges faced by signal detection in complex electromagnetic environment. Main research results are as follows:In weak signal detection, we introduce the nonlinear stochastic resonance(SR) and use the noise to enhance the weak signal. The optimal bistable stochastic resonance system were first designed, and then applied in the big parameters by using the linear transformation method. We designed the Bistable stochastic resonance based Energy Detection(BED) method. The BED enhances the signal by using the noise, and performs better than traditional ED. We further proposed the Noise Enhance Energy Detection(NEED) method by the generalized stochastic resonance. For the proposed algorithm, suitable SR noise is added to modify the probability distribution of the detection statistics. Introducing the deviation coefficient, the NEED can get the optimal SR noise, so that enhance the traditional ED.Signal detection in the presence of non-Gaussian noise is a challenging problem. However, there are few detectors that can work well in this case. We propose a signal detection algorithm via absolute value cumulating(AVC) with Laplacian noise. A performance analysis about the influence of noise uncertainty in the low signal-to-noise ratio regime isalso given, which shows that the SNR Wall of the AVC is half of that of the energy detection. The AVC algorithm is further introduced into existing cooperative signal detection scheme. Simulation results validate the algorithm, and show that the proposed algorithm can improve the performance of existing algorithm at least 3 d B with Laplacian noise. In order to fit the high dynamic noise and overcome the noise uncertainty, a Goodness of fit based method is proposed with Laplacian noise. The proposed algorithm performs well and has the advantage of invulnerability to the noise uncertainty. In generalized Gaussian noise, Frequency domain Goodness of Fit Test(FGo F), is proposed. The FGo F makes full use of underlying information in Guard-Bands, and the inherent advantages of Go F test works for any distribution. The proposed method has the inherent advantages of invulnerability to noise uncertainty and noise model error.In wideband multiple signal detection, a performance analysis of the frequency domain energy detection based on periodogram method is provide. We further expand the FGo F method into wideband multiple signal detection. Analytical and simulated results show that the FGo F is a robust signal detection method with the inherent advantages of invulnerability to dynamically varying noise, and suitability for wideband signal detection.In the research of quick signal detection, we first propose a signal method based on double sensing length. For the proposed algorithm works more efficient than conventional energy detection, meanwhile it possesses the advantages of less average sample number(ASN) and lower complexity. We further propose a Generalized stochastic resonance based Segmental Energy Detection(GSED) method. Both of the simulation results and theory results show that the GSED can shorten the ASN further, compared to the previous method. To resolve the problem of high overhead of cooperative sequential detection, a Cooperative detection method by Dual Sequential Detection(CDSD) is proposed. For the proposed method, a sequential probability ratio test(SPRT) is taken in both each sensor and fusion center, which makes the reporting messages more effective. Both of the simulation results and theory results show that the CDSD algorithm has the comparable ASN with much low cooperative overhead.