| Individual identification of radiation source,also known as Specific Emitter Identification(SEI),or “fingerprint” identification,refers to the process of determining the individual radiation source that generates the signal based on certain criteria or prior knowledge by measuring the received electromagnetic signal.In the military field,this technology could achieve accurate tracking and attack.It is an important foundation for high-level information fusion,and also is an important support for situation estimation.In the civilian sector,it can be used in wireless access authentication,electromagnetic spectrum monitoring and some other public safety projects.At present,domestic research on this technology is often performed well in laboratory environment due to high equipment cost and unstable algorithm.But in actual non-ideal signal environment,the existence of noise and multipath interference leads to instability of individual characteristics,the performance of the classifier decline and failing to reach the recognition requirements.Aiming at these problems,this paper focuses on the aspects of individual feature extraction,information fusion model improvement and minimum verification system design.The main contents are as follows:1.According to the actual signal environment,the commonly used single-channel and I/Q dual-channel receiving equipment,two kinds of received signal models are established.The individual parameter estimation methods of typical pulse signals are summarized and simulated.The advantages of the algorithm are summarized and the limitations of the algorithm are also pointed out.2.Aiming at the problems of high spatial characteristics,redundant information and high noise of conventional radar radiation sources,the individual feature extraction algorithm of radiation source based on Bezier curve fitting is studied.The method uses the Bezier curve to fit the conventional features,and the fitted control points are taken as the individual feature vectors.Experiments show that compared with the traditional individual identification method,the feature dimensions extracted by this paper is lower,the storage space is smaller,the recognition efficiency is higher,and it is suitable for large-scale digital processing without significantly reducing the recognition rate.3.Aiming at the fact that the conventional single classifier is vulnerable to the change of the external environment in practical applications,and the output of the information fusion model based on decision-making layer is not ideal,this paper proposes an improved recursive centralized spatiotemporal information fusion model based on Dempster-Shafer(DS)evidence theory.The model inherits the advantages of the traditional recursive centralized information fusion model,and at the same time effectively reduces the failure of individual classifiers in recognition process and the adverse effects of the fusion order on the model output,enhances the support for the true target and relatively reduces the support for other false target individuals.The experiments show that the proposed model has good real-time performance and stable performance.4.For the current domestic research,the individual identification technology of radiation source is mostly based on theoretical simulation and lack of verification equipment for field measurement.A minimum verification system for individual identification of radiation source based on FPGA is designed.The system could realize collection of the ADS-B signals,storage,decoding,individual feature extraction,database access and individual identification.Finally,through the field test,the average individual recognition accuracy reached 88.3%. |
