Research And Implementation Of Wireless Communication Interference Signal Cognition Technology

Wireless communication,as a basic technology involved in people’s daily life,has achieved rapid development in recent years,from 4G to 5G,bringing great convenience to the world,but also making the electromagnetic environment in space increasingly complex.In addition,communication signals in space can also be subject to various types of interference,especially under malicious human interference,and communication systems may even face the risk of collapse.Moreover,traditional cognitive algorithms also face problems such as high computational complexity and difficulty in deploying on devices with limited resources.Therefore,anti-interference technologies such as interference recognition with low complexity are particularly important.Especially in today’s severe international situation,adopting effective anti-interference technology in military warfare can also ensure the normal operation of our communication equipment.Therefore,this article focuses on researching efficient interference detection and recognition algorithms,which greatly reduces the complexity of hardware implementation while ensuring the accuracy of cognition.Finally,an efficient interference recognition algorithm is designed,and the FPGA implementation framework of the interference recognition algorithm is emphatically studied.Finally,based on the hardware platform,the performance of interference recognition is measured using actual interference signals,verifying the correctness of the algorithm.Firstly,this article designs a wideband interference signal recognition scheme.Adopting the scheme of multi subband interference cognition,the wideband interference cognition was divided into multiple narrowband interference cognition,and time was exchanged for resources and accuracy.Finally,a data structure for multi subband interference cognition results and a comprehensive strategy for multi subband interference cognition results were designed.And a database of interference signals,including simulation data and actual collected data,was constructed for the interference signal model.Secondly,this paper studies the technology of wireless communication interference signal recognition.Based on the shortcomings of traditional interference detection and recognition,it proposes interference recognition algorithms based on artificial feature extraction depth neural network,interference recognition algorithms based on convolutional neural network automatic feature extraction,interference recognition algorithms based on adaptive forward propagation convolutional neural network,and interference recognition fusion algorithms combined with multi domain features,respectively,A total of 5 classifiers were designed,including MFE-DNN interference classifier,CNN-WSS interference classifier,CNN-TFI interference classifier,ACNN-WSS interference classifier,and CNN-JMDF interference classifier.The recognition performance of different classifiers was compared.Thirdly,this paper designs an FPGA implementation scheme for interference recognition,and designs implementation schemes for standardized processing,Welch spectrum estimation,and time-frequency image calculation for preprocessing.For CNN networks,it designs implementation schemes for CNN-WSS interference recognition networks,CNN-TFI interference recognition networks,and ACNN-WSS interference recognition networks.It also designs a detailed implementation scheme for general modules in CNN networks.Fourthly,this article established a testing system and developed a testing interface.Firstly,the performance of interference detection was tested.The mean square error of the measured and theoretical values can meet the accuracy requirements under a 50 MHz analysis bandwidth.Then,the recognition performance of the CNN-WSS and ACNN-WSS interference recognition networks was tested separately.After fine-tuning the actual data,the recognition rate of the CNN-WSS network can reach 100% when the dry to noise ratio is 0d B,The ACNN-WSS network can achieve good recognition rate while increasing the average recognition speed by 190%.