Research On Channel Estimation In RIS Assisted Wireless Communication System

Reconfigurable Intelligent Surface(RIS)technology,as a new communication technology,can artificially improve the communication environment in the process of wireless communication by arranging a large number of RIS passive components,and can obtain high communication quality at a very low hardware cost.It is recognized as one of the most promising communication technologies in the future.The communication channel of RIS auxiliary wireless communication system is composed of direct channel and cascaded channel.The direct channel can be estimated by conventional channel estimation algorithm,but there are two difficulties in the estimation of cascade channel: firstly,the components of RIS are passive devices,which cannot transmit,receive or process pilot signals;Secondly,RIS is generally composed of a large number of reflection units,resulting in a much larger dimension of the cascade channel than the conventional channel.Since then,the problem of cascading channel estimation has become the research focus of RIS-assisted mobile communication system.Compressive Sensing(CS)technology,as a new signal processing technology,can collect and compress data at the same time.Its signal sampling rate is far lower than the Nyquist sampling rate,which greatly saves the transmission cost and solves the problem of signal processing with large amount of data.CS technology is mainly used to acquire,process and recover sparse signals,and is often used in channel estimation of mobile communication systems.This paper focuses on the sparsity adaptive channel estimation method for the cascading channel of RIS-assisted wireless communication system based on compressed sensing theory,and studies the influence of the design of RIS reflection coefficient on the performance of the cascade channel sparsity adaptive estimation algorithm.The main research contents and work of this paper are as follows:(1)Through studying the basic ideas and methods of a large number of sparse adaptive channel estimation algorithms,based on the Double-Structured Orthogonal Matching Pursuit(DS-OMP),an adaptive Double-Structured Orthogonal Matching Pursuit(ADS-OMP)for the channel estimation in the RIS-assisted wireless communication system is proposed.ADS-OMP algorithm can realize high-quality reconstruction of cascaded channels without knowing the sparse information such as row sparsity,column sparsity and common column sparsity of the joint channels in the angular domain.It solves the problem that DS-OMP algorithm relies excessively on the sparse information of the concatenated channels in the diagonal domain,and realizes the sparsity adaptive channel estimation of the joint channels.Simulation results show that under different signal-to-noise ratios,the normalized mean square error(NMSE)performance of ADS-OMP algorithm is highly consistent with the performance of DS-OMP algorithm,and they have approximately the same channel recovery ability.On the other hand,the complexity of ADS-OMP algorithm is slightly higher than that of DS-OMP algorithm,but the ADS-OMP algorithm does not rely on the relevant sparse information related to the concatenated channels,and the algorithm is more practical.(2)Through the research and learning of a large number of classic compression sensing algorithms,and further improving the ADS-OMP algorithm,an adaptive double-structured sparse generalized orthogonal matching pursuit algorithm(ADS-GOMP)based on generalized orthogonal matching pursuit algorithm(GOMP)is proposed.ADS-GOMP mainly improves the atomic screening method,which is different from the ADS-OMP algorithm that only filters one atom at a time.This algorithm combines the idea of GOMP algorithm to adjust the number of atomic screening according to the signal sparsity.At the final stage of the algorithm,the number of atomic screening is dynamically adjusted according to the estimated information such as row sparsity and column sparsity.In the face of signals with large sparsity,it has a faster convergence speed than ADS-OMP algorithm.Simulation results show that at low SNR,ADS-GOMP algorithm has the channel estimation ability similar to that of ADS-OMP algorithm.When the number of users is 32,the complexity of the algorithm can be reduced by more than 20% as compared to ADS-OMP,and it will be further reduced with the increase of the sparsity of the angular concatenated channel.With the improvement of signal-to-noise ratio,there will be a small performance gap between ADS-GOMP and ADS-OMP algorithm,but not more than 1d B.(3)The influence of the design of RIS reflection coefficient on the estimation performance of ADS-OMP algorithm and ADS-GOMP algorithm is studied.In the framework of channel estimation based on compressed sensing,the reflection coefficient matrix corresponds to the equivalent measurement matrix one by one.The design of the reflection coefficient of the RIS system can be transformed into the optimization of the equivalent measurement matrix in the theory of compressed sensing.Through learning a large number of measurement matrix optimization algorithms,the measurement matrix optimization algorithm based on differential evolution algorithm is studied,and the selection method of population is improved to further improve the optimization effect.Simulation results show that the improved measurement matrix optimization algorithm can obtain the measurement matrix with small mutual coherence,and on this basis,the optimized equivalent measurement matrix can be obtained,from which the corresponding reflection coefficient matrix can be obtained and the reflection coefficient design can be completed.The equivalent measurement matrix before and after optimization is simulated with ADS-OMP algorithm and ADS-GOMP algorithm respectively.The results show that the optimization of the equivalent measurement matrix can improve the channel estimation performance of the ADS-OMP algorithm and ADS-GOMP algorithm,especially at low signal-to-noise ratio,with a performance gain about 1-2d B.