Electromagnetic signal intelligent cooperative identification method and system

Abstract

The invention belongs to the technical field of electromagnetic signal intelligent identification, and discloses an electromagnetic signal intelligent cooperative identification method and system, andthe method comprises the steps: carrying out the intelligent representation of an electromagnetic signal, and taking the electromagnetic signal as the input of a subsequent deep learning network; constructing a DenseNet-based feature fusion network on each distributed sensor to perform feature-level fusion on the intelligent representation; fusing the loss functions obtained by network training by adopting a federated learning network architecture and feeding back the fused loss functions to each DenseNet network for training; and implementing electromagnetic signal identification based on distributed decision-making level fusion. When the generalized signal-to-noise ratio is more than 10dB, the recognition rates of AM, FM, BPSK, QPSK, 8PSK, 2ASK, 4ASK, 2FSK and 4FSK signals are all morethan 90%, so that the method and system has good recognition performance.

Description

technical field [0001] The invention belongs to the technical field of electromagnetic signal intelligent identification, and in particular relates to an electromagnetic signal intelligent collaborative identification method, system, storage medium and application. Background technique [0002] At present, electromagnetic signal classification is a basic and key technology in electromagnetic signal cognition. It is used to identify the modulation information of electromagnetic signals and has a wide range of applications in signal demodulation, suspicious transmission monitoring, anomaly detection, and interference location. . In a distributed network, due to the influence of different propagation and transmission environments, even if the transmitter sends out the same signal, different signals will be observed on each receiving sensor. Traditional signal recognition methods mostly use Gaussian noise as the noise model, but the noise in the actual environment often has the...

AI Technical Summary

Problems solved by technology

However, this method cannot guarantee the global optimality of the model (R.Cao, J.Cao, J.-P.Mei, C.Yin, and X.Huang, ``Radaremitter identification with bispectrum and hierarchical extreme learning machine, "Multimedia Tools Appl., vol.77, pp.1_18, May 2018.)

G.Yang et al. proposed a method based on explicit infinite cost function and global optimization, but when the oversampling rate is reduced, the performance of the algorithm will decrease (G.Yang, J.Wang, G.Zhang, Q.Shao and S .Li, "Joint Estimation of Timing and Carrier Phase Offsets for MSK Signals in Alpha-Stable Noise," IEEE Communications Letters, vol.22, no.1, pp.89-92, Jan.2018.)

M.Mohanty et al. proposed a modulation type classification method after using a complete mixture dictionary to perform sparse signal decomposition (SSD) on additive mixed Gaussian noise and impulse noise. However, this method requires SSD to perform additional pre-processing on the received signal. processing, thereby increasing the computational complexity of the classifier (M.Mohanty, U.Satija and B.Ramkumar, "Sparse decomposition framework for maximum likelihood classification under alpha-stable noise," 2015 IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT), Bangalore, 2015, pp.1-6.)

However, the above classification effects either choose an algorithm that may lead to slow convergence and cannot guarantee the global optimality of the model because of performance considerations, or because a large amount of preprocessing of the signal increases the complexity, or because the signal quality is high. When the environment changes, good robustness cannot be guaranteed, and the ideal effect cannot be achieved.

Images