Target recognition for satellite communication by employing higher-order statistics

The target recognition for satellite communication (satcom) is generally regarded as the cutting edge of electronic countermeasure research. This work is dedicated to the investigation on the theory and experiment of satellite communication target recognition on the basis of systematical analysis of satcom signal emission, propagation, and reception. The authors elaborate on the fingerprint analysis, feature extraction, and identification of satcom emitters by utilizing the nonlinearities of high-power microwave vacuum amplifier (HPA). The mechanism of the external subtle features of satcom signal will be also discussed in detail. To acquire the qualified features that precisely represent the individual emitter, higher-order statistics technique is introduced to implement the feature extraction, and the supervised probabilistic neural network classifier is established to execute the recognition of testing satcom samples. In testing phase, there are a total of 4000 sampling signals with BPSK modulation and variable carrier to noise ratio (CNR) originated by eight types of satcom transmitters setting for the experiment to verify the authors' viewpoints. Thanks to the fine training data set and subsequent well-extracted features, the PNN classifier had not fail us and finally achieved satisfactory accuracy of more than 94% at CNR level of 10 dB. Those expected results will help to enhance the ability of battlefield surveillance and situational awareness that is of paramount importance in academic research and military application.