A Research On DOA Estimation Algorithm For Wideband Signal In Massive MIMO System

Direction of arrival(Direction of Arrival,DOA)estimation is widely used in radar,communication,electronic reconnaissance and so on.It is the most important research hotspot in array signal processing.For the wireless mobile communication system,the effective location information of the user is very important to the accurate and reliable transmission of the information,and the DOA estimation can realize the angle parameter estimation of the direction of arrival of the user and give out the specific location information.At present,massive MIMO technology has been applied to 5G mobile communication because of its outstanding advantages,which brings higher transmission rate and improves the capacity of the communication system.In addition,the technology effectively reduces interference,enhances coverage,and makes the whole network more flexible.Therefore,massive MIMO technology has become the main research area of current scholars.Meanwhile,the DOA estimation is of great importance to the communication system because of its many advantages.Wideband signal is different from narrow-band signal,it has strong anti-jamming ability,carries more information,and has less correlation with noise.Therefore,the wideband signal has a more extensive application in the communication.However,wideband is more complicated than narrow-band signal,so it is difficult to realize in real-time.So in this paper,DOA estimation for wideband signals will be studied.While solving the problem of high complexity of wideband DOA estimation,it will be better applied to massive MIMO systems.Wideband signal DOA estimation algorithms are mainly divided into two categories,one is incoherent signal subspace estimation algorithm(ISM),the other is coherent signal subspace estimation algorithm(CSM).Compared with the ISM algorithms,the CSM algorithms has the advantages of less computation,higher resolution,and the most important one is that it can estimate the coherent signal source.Therefore,based on CSM algorithms,this paper focuses on the methods and criteria of constructing the focusing matrix.Aiming at the problem that most coherent signal subspace estimation algorithms need angle estimation,a new focusing matrix method is proposed.This method does not need angle estimation,and has a small amount of computation.In order to further solve the problem of high computational complexity and poor real-time performance of the algorithm in massive MIMO systems,an improved fast wideband DOA estimation method based on PCA neural network is proposed in this paper.The PCA neural network does not need prior sample training or eigenvalue decomposition of array covariance matrix when estimating signal subspace.It only needs a limited number of self-organizing learning to estimate the weights of the network and then get the signal subspace.Therefore,the proposed algorithm has low computational complexity and strong timeliness,and is suitable for massive MIMOsystems.Then,the AIC criterion is used to provide more accurate subspace dimension for PCA neural network signal subspace estimation,which makes the proposed algorithm have better performance.Through a large number of simulation experiments,it can be obtained that the improved algorithm can still accurately estimate the angle of arrival of the signal source under the condition of small snapshot number and low signal-to-noise ratio(SNR).