Design Of Wireless Communication Signal Classification Algorithm Based On Deep Neural Network Under Additive White Gaussian Noise

In the past few decades,the number of wireless communication devices and the amount of data exchanged between them has continued to grow,leading to exponential growth in demand for wireless communication resources by many users.Wireless communication signals have become an important research direction.Due to the recent research boom in machine learning and deep learning,this article focuses on combining deep learning with wireless communication signal recognition.The performance in the field of automatic modulation signal identification and interference signal identification in unlicensed bands are under researching respectively.The main work objectives and innovative content of this article is shown blow:1.Automatic Modulation Recognition(AMR)technology has many potential applications in the field of modern wireless communication field,especially in cognitive radio who relies on AMR to respond to changes in the surrounding environment and adjust the strategy simultaneously.Generally speaking,the device signal receiver needs to receive the signal accurately.Therefore,in the process of its design,being able to effectively identify the modulated signal is one of the most important indicators for measuring a device signal receiver.When the signal receiver can correctly identify the modulation method of an unknown signal,we can correctly demodulate the signal according to its modulation method,and can also accurately estimate the frequency and bandwidth information of the signal.In recent years,reliable AMR methods based on deep learning have been developed.However,all of their AMR training models are considered in specialized channels rather than generalized channels.Therefore,these AMR methods are difficult to apply in general situations.In this article,we propose a deep learning-based blind channel recognition assisted AMR method,which is performed by two independent convolutional neural networks.The first convolutional neural network was trained on in-phase and quadrature sampled signals to distinguish the channel conditions of the received signal.The second convolutional neural network is trained under line-of-sight channel and non-line-of-sight channel conditions.The simulation results show that the recognition accuracy of our proposed AMR system assisted by blind channel recognition is significantly better than that of the traditional AMR system.2.With the rapid development of Long Term Evolution(LTE)technology,LTE technology in unlicensed band(LTE-U)can effectively solve the problem of insufficient spectrum resources.This is why the major operators are eager to find a network technology that can meet today's increasing traffic demand and can make better use of the licensed spectrum.As early as 2014,the 295 MHz frequency band in the 5 GHz spectrum was opened for unlicensed bands,which inspired a new round of research activities: the use of unlicensed bands together with licensed bands was the LTE network at that time Bring new vitality in the architecture.However,the competition between LTE-U and Wi Fi will seriously affect the communication quality of them,which makes the friendly coexistence of LTE-U and Wi Fi being a research focus.According to this hot issue,this paper proposes a deep learning-based Wi Fi and LTE-U signal recognition algorithm,especially a classification algorithm based on convolutional neural networks to realize the recognition of LTE-U and Wi Fi signals.The experimental input uses mixed data under different signal-to-noise ratios,and compares the classification results in the two data forms.The experimental results show that the algorithm based on the convolutional neural network model proposed in this paper can effectively distinguish LTE-U and Wi Fi signals,and further realize the friendly coexistence of them.