Research On Signal Separation Method For Time-frequency Overlapped Digital Communication Signal From Single Antenna

In nowadays information era, the types of communication or radar electronic equipment for both military and civilian applications are increasing significantly, which eads to various communication and radar signals overcrowded in time domain, overlapped in frequency domain and intertwined in space domain shaping a more complicated electromagnetic environment. As a result, the interception probability of ime-frequency overlapped signals has been improved for communication econnaissance equipment. In order to accurately capture the interested signal or liscover the interference signal, separating these time-frequency overlapped signals and xtracting the information implied in the useful signal have become a research task with significant for electromagnetic surveillance domain. Meanwhile, the communication econnaissance equipment with single antenna not only reduced the cost but can reduce he equipment volume and improve its flexibility, taking this background in to consideration, this dissertation focuses on time-frequency overlapped signal separation lgorithms for single-antenna communication reconnaissance.Main contributions of this dissertation are listed as follows:1. Based on dynamical embedding framework, a single antenna separation lgorithm is proposed for time-frequency overlapped digital communication signal, irstly, the dynamical embedding model is built, based on which the single-antenna ecording signal is projected into multi-dimensional virtual measurement matrix. The nodel parameter setting principle is proposed based on the feature of multi-dimensional neasurement matrix under over-determined or determined condition, and proved that he dynamical embedding model can satisfy the prerequisites of independent component ilgorithm. After that, independent component analysis algorithm is employed to process he obtained virtual matrix in order to extract independent components, which are rojected back into original measurement space based on the dynamical embedding nodel. At last, the estimated components of original sources are reconstructed and elected based on the frequency spectrum feature of the obtained independent components and the inherent independent characteristic among the original sources, Moreover, this algorithm is also suitable for undetermined signal separation, which extends the independent component analysis algorithm’s application. Numerical simulation results verified the effectiveness of the proposed algorithm.2. A signal separation algorithm for time-frequency overlapped digital communication signal from single antenna is proposed based on multi-scale wavelet lecomposition. Firstly, wavelet scale is set according to frequency band of mixed signal eceived by single antenna communication reconnaissance equipment. And then, nulti-scale wavelet decomposition is employed to extract the different frequency band signal components of mixed signal, and those signal components are used to build the nulti-dimensional virtual measurement matrix. After that, independent component inalysis algorithm is adopted to extract the independent components and the estimated components of original sources are selected based on the frequency spectrum feature of he obtained independent components. The feasibility of the proposed method is proved )y theoretical analysis and verified by numerical simulation.3. Considering multiple time-frequency overlapped binary phase shift keying signals are intercepted by single antenna communication reconnaissance equipment, requency estimation method is studied based on second order cyclic cumulant, which lot only avoid the cross-component influence on original signal frequency estimation ind provide the original source number estimation. Then, according to original signal )arameter estimation, and using blind-adaptive Frequency-Shift filter of the cyclic requency domain, a single antenna signal separation method for time-frequency overlapped is proposed based on signal cyclostationary by building dual signal model of nultiple time-frequency overlapped signal received by single antenna. The feasibility of he proposed method is proved through simulation.4. Two signal separation algorithms for time-frequency overlapped signal from single antenna are proposed based on the dissimilarity among the original sources. One s based on the distinct feature to different carrier frequencies among original sources ind the other one is based on different types of mixed time-frequency signals. The first eparation method employs Short-Time Fourier Transform to extract the amplitude and )hase of each original source, and using the extraction information to reconstruct riginal signals. This method is simple and still effective. Under the complex and varying electromagnetic environment, a problem is that communication signal may be nixed by a time-frequency overlapped radar signal with fixed pulse reception interval. To overcome this problem, a signal separation approach is proposed based on the signal distinction for different types. Usually the power of communication signal is much smaller than that of radar signal, hence the pulse repetition interval of radar signal can )e estimated. Then, the mixed signal is separated by using the periodic characteristics of adar signal and singular value decomposition technique simultaneously. The second nethod has less requirements on communication signal modulation type and release any requirements on radar signal’s intra-pulse information.