Specific radiation source identification method based on continuous learning and joint feature extraction

Abstract

The invention discloses a specific radiation source identification method based on continuous learning and joint feature extraction, and the method comprises the steps: obtaining signals of a plurality of radiation sources, and carrying out the signal processing of the signals of the plurality of radiation sources; inputting the signals of the plurality of radiation sources after signal processing into the plurality of trained continuous increment depth extreme learning machines, and taking the plurality of trained continuous increment depth extreme learning machines as classifiers to output decisions; and fusing the output decisions of the single continuous increment depth extreme learning machine by using a voting algorithm, and selecting the class with the highest confidence coefficient as a classification result to identify a specific radiation source. According to the method, the recognition precision is still high under the condition of a small number of samples, continuous supervised recognition of the collected samples can be achieved, the requirement for dynamic updating of a database is effectively met, good compatibility can be shown for different modulation modes, carrier frequencies and receiving and transmitting distances, and effective recognition can be conducted on multiple emitters.

Description

technical field [0001] The present invention relates to the field of communication technologies, in particular to a method and device for identifying a specific radiation source based on continuous learning and joint feature extraction. Background technique [0002] Specific emitter identification (SEI) refers to a technology that uses the inherent defects of the physical layer of hardware devices to identify a single emitter. It is widely used in spectrum management, cognitive radio, and self-organizing networks. In real channels, the additional nonlinear distortion of the ji intercepted signal is often unreproducible, so it is feasible to use the radio frequency fingerprint (Radio Frequency Fingerprint, RFF) feature to determine the device label. The supervised SEI based on RFF extraction is often divided into two stages: the first stage is feature extraction based on transient signals or steady state signals, and the second stage is to build a classifier to train and disc...

AI Technical Summary

Problems solved by technology

However, the existing DL-based SEI models are usually established on data sets with sufficient samples and complete labels, and in actual non-cooperative communication scenarios, the sample size is often limited; and the database is changing dynamically, and the training of DL models is usually It is a single learning based on existing samples. Once the training set changes, it needs to be retrained. The self-updating ability of the parameters of the model is poor.

Images