Research On New Channel Estimation Technology For 5G

In the vision of 5G,people hope that anyone and anything can obtain and share information at any time in any place in the future.With the rapid development of the data age,people's requirements for the transmission speed,capacity,and spectrum efficiency of information communication are constantly increasing.Massive MIMO is a technology that realizes people's needs and brings significant changes.As researchers continue to explore and improve,massive MIMO has achieved remarkable results in both conceptual standards and practical applications.As a relatively mature modulation technology,OFDM,which will be more conducive to high-quality wireless communication.In order to obtain a better communication transmission effect,channel state information is essential for reliable communication.At the receiving side of the wireless communication system,the channel state information will be used for coherent detection and decoding.Therefore,it is extremely important to use a channel estimation algorithm with high accuracy and low computational complexity.Massive MIMO channel estimation algorithms are divided into three categories: pilot-based nonblind,semi-blind and blind channel estimation.In this paper,through the derivation of the pilot-based non-blind and semi-blind channel estimation formula and the error analysis of the corresponding algorithm,the simulation compares performance of the non-blind and semi-blind channel estimation,and innovatively analyzed and compared the pros and cons of the computational complexity of the two semi-blind channel estimation algorithms in massive MIMO systems under different conditions,and provided suggestions on the computational complexity of the choice of channel estimation algorithms under actual conditions.In recent years,with the continuous upsurge of artificial intelligence,the original theoretical analysis may not be applicable to complex wireless communication system.Naturally,researchers have cast their sights on the possibility of applying deep learning to channel estimation.Through the training of a large amount of data,this new type of channel estimation method,through training and learning of a large amount of data,analyzes the characteristics of the channel itself and finally forms an optimal model to complete the channel estimation task autonomously.Through the analysis of the existing channel estimation schemes based on deep learning,this paper innovatively classifies and summarizes the characteristics of channel estimation based on deep learning from the perspective of model structure and input and output data set selection and puts forward own views.