Research On Large-scale MIMO Semi-blind Channel Estimation And Space-time Coding Based On Tensor Decomposition

Massive multiple input multiple output(Massive MIMO)technology has greatly improved the performance of the system by adding a large number of antennas and has become one of the key technologies of the fifth-generation mobile communication technology.With the increase of antenna number,the dimension of channel matrix is improved,which leads to the complexity of channel estimation and signal detection.It is necessary to find an appropriate channel and signal estimation method.If pilots are used for channel estimation,although they are commonly used and easy to implement,massive MIMO requires the use of a large number of pilots,which will occupy a lot of system frequency band resources,reduce system performance.Moreover,there is a hidden danger of pilot pollution,which reduces the accuracy of channel estimation.Signal detection techniques that require known channel state information(CSI)are largely affected by the accuracy of the system's channel estimation.In order to reduce the influence of pilots on channel estimation and the influence of channel estimation accuracy on signal detection,a tensor-based multi-dimensional signal processing technology has emerged.By making full use of the multi-dimensional structural attributes of channel and signal presentation,the effects of reducing pilot overhead and improving the performance of channel estimation and signal detection can be achieved.In this paper,we study the problem of channel estimation in massive MIMO system and signal detection in space-time code by tensor decomposition.The received multidimensional signals are constructed as tensor models and then the low-rank decomposition is carried out.Under the premise of satisfying the unique decomposition,the parameters of the model are estimated by using appropriate fitting algorithm.The main work of this paper is as follows:1.Considering the influence of reducing pilot overhead,this paper makes use of the multi-dimensional structure information characteristics of channel and signal to conduct tensor modeling and analysis,and on this basis,comprehensively uses the multi-dimensional signal processing method.Through joint processing,the high dimensional structural information is fully mined and the multi-dimensional channel and signal structural information is effectively used for channel estimation and signal detection.On the premise of satisfying the unique decomposition of the highdimensional tensor,an appropriate fitting algorithm is used to solve the problem,so as to obtain satisfactory results of channel estimation and signal detection,and at the same time to achieve the effect of reducing system pilot overhead.2.For channel estimation in massive MIMO,the parallel factor(PARAFAC)decomposition iteration fitting is used to complete the channel estimation by taking full advantage of the multi-dimensional characteristics of the received signal.According to the structural characteristics of the constant mode of the sending signal in the communication signal,the received signal tensor model is constructed as the structural constraint PARAFAC decomposition.According to the projective alternating least squares(PALS)fitting method,the source is mapped to the constant modulus data set in the iteration process and then the source is reconstructed to complete the channel estimation more accurately.The simulation results show that,with the aid of a small amount of pilots,the algorithm used effectively improves the channel estimation accuracy and saves the frequency band resources of the system.3.Considering the characteristics of space-time structure of transmission signals,this paper combines tensor with linear space-time Code and proposes multi-block space-time Code(MSTC),which extends sending characters into multi-block spacetime data sub-streams.Block-PARAFAC decomposition is used for blind signal detection and the standard alternating least square algorithm(ALS)is used to iteratively fit and solve the model.In the case of unknown channel status,the transmitted signal can be effectively estimated.The simulation results show that the proposed algorithm can realize blind detection of the receiving sub-stream without the need of pilots and achieve a lower bit error rate when the signal-to-noise ratio is low,which improves the system performance and has a higher spectral efficiency.