Wireless transmitter recognition method based on multi-core double-channel network

Abstract

The invention discloses a wireless transmitter recognition method based on a multi-core double-channel network, and belongs to the field of electromagnetism. The method specifically comprises the following steps: firstly, performing power amplification on a modulation signal through a transmitter, transmitting the modulation signal to a receiver for sampling and receiving, performing statistical characteristic processing on a received signal to obtain a cyclic spectrum, and extracting a high peak density value in the cyclic spectrum as a characteristic; then, inputting all the characteristic values into a fully connected multi-core double-channel network module, learning the characterization characteristics of each transmitter from the cyclic spectrum, and fusing the obtained result unionsets; after a multi-core double-path network model is trained, the fused feature result is tested, and normalized feature vectors corresponding to all test features are obtained; and obtaining a probability value corresponding to each feature through a softmax classifier, and selecting the wireless transmitter with the maximum probability value as the final wireless transmitter to be identified and classified. According to the method, the calculation complexity is simplified, and the situation that the classification effect is not ideal due to noise interference is reduced.

Description

technical field [0001] The invention belongs to the electromagnetic field and relates to microwave radio frequency technology, in particular to a wireless transmitter identification method based on a multi-core dual-channel network Background technique [0002] In terms of wireless communication networks, there are a large number of illegal users involved, and the identity marks of the software layer are easily tampered with, making it difficult to identify the access of some illegal devices. At present, the traditional network security and other methods further lock the user's identity, which can greatly improve the security of the communication network; therefore, the wireless transmitter identification technology based on the inherent attributes of the hardware layer of the access device transmitter came into being. This technology uses methods such as cyclic spectrum and neural network to analyze the nonlinear characteristics of the wireless transmitter on the hardware, ...

AI Technical Summary

Problems solved by technology

[0006] However, the above document simulates different transmitters based on adding phase noise to the signal at different signal-to-noise ratios, extracts features based on bispectral analysis, uses convolutional neural network to train it to identify different transmitters, and adds phase to the signal The noise method simulates different transmitter signals, and cannot describe the collection of wireless transmitters more comprehensively, resulting in limited actual recognition results.

Images