Research On Spectrum Cognition Method For Satellites Communication System

In emergency communication scenarios,the ground base stations in the disaster area are destroyed,which makes it difficult to provide timely and reliable communication.Satellite communication has the advantages of being flexible and free from distance limitation,and commonly serves as an essential guarantee for emergency communication.Therefore,in emergency communication scenario,accurate cognition of the spectrum of satellite communication system is the key to ensure the utilization efficiency of communication resources and maintain emergency communication.Satellite signals are characterized by long transmission distance,high transmission delay and low signal-to-noise ratio,which brings great challenges to satellite signal cognition.Only if fast and accurate spectrum cognition is realized,can the current state of satellite transmission be recognized and used effectively before it changes.Therefore,considering the characteristics of long transmission distance and low signal-to-noise ratio of satellite signals,this thesis proposes a spectrum cognition scheme for satellite communication system to improve the accuracy and timeliness of cognition.The research contents of this thesis are as follows:First,satellites signal has a long transmission distance and a low signal-to-noise ratio,which poses a challenge to the spectral sensing scheme for rapid sensing under low signal-to-noise ratio.In view of this challenge,this thesis proposes a spectrum sensing scheme based on XGBoost model using signal phase difference features:Firstly,the training data set is constructed by extracting the phase difference distribution of sampling signal and noise.Secondly,combined with the XGBoost algorithm,the spectrum sensing problem is modeled as a binary classification problem.Then,the XGBoost model is trained based on the training data set to realize spectrum sensing under different false alarm probability.Simulation results show that when the SNR is-15dB,the performance of the proposed scheme is 14%higher than that of the spectrum sensing scheme based on DNN,and the training time of the proposed scheme is less than that of the scheme based on DNN.Second,on the premise of sensing the signal,in order to help determine the source of the signal and identify the type of satellite,this thesis further identifies the signal modulation mode,and proposes a modulation recognition scheme based on the combination of multi-input CNN and LSTM.The proposed scheme models modulation recognition as a pattern recognition problem,by inputting multi-dimension signal information,CNN and LSTM are used to realize feature extraction,and a deep learning model is established to realize modulation recognition.The simulation results show that the proposed scheme is superior to the MCLDNN model(the best performance in current literatures),and the recognition accuracy increases by 7%in a short sampling time,which is more in line with the"fast and accurate" requirements of spectrum cognition in emergency communication scenarios.