Research On Sparse Precoding Matrix In Beamspace For Massive MIMO Systems

In downlink massive multi-user MIMO systems,the base station transmits independent data streams to multiple users on the same time-frequency resources,which leads to a multi-user interference(MUI)problem,and this problem can be effectively solved by precoding techniques.The implementation of precoding techniques in real communication systems faces many difficulties,such as high computing load and large storage overhead.It is worth noting that in a finite scattering environment,the user’s channel model usually exhibits structured sparse properties in the beam domain.We can take advantage of the sparsity of the beam-domain channel matrix to design a sparse precoding matrix,thereby reducing the complexity of precoding symbol processing and storage overhead.The main works in this thesis are as follows:(1)To deal with the problem of high complexity of precoding matrix design in the existing sparse precoding algorithm,we propose a beam-domain sparse precoding matrix design scheme based on the maximum energy criterion.The scheme adopts beam selection based on the maximum energy criterion to ensure the sparsity the the sparsity of the precoded matrix and maximize the channel diversity gain brought by multiple antennas.The analysis and simulation results show that,compared to the existing sparse recovery-based sparse precoding matrix design scheme,the proposed algorithm has better performance and lower computational complexity.(2)Aiming at the problem of low reachable rate of the existing beam-domain sparse precoding algorithm,a sparse precoding matrix design scheme to maximize the reachable rate is proposed.The scheme converts the total achievable rate optimization problem with non-convex constraints into a series of simple sub-optimization problems,and obtains the precoding matrix column by column with successively solving the suboptimization problem,while using the Aitken-accelerated power algorithm to solve the main eigenvalues and corresponding eigenvectors of the matrix to avoid the complex large-dimensional matrix SVD decomposition to reduce complexity.The analysis and simulation results show that the achievable rate of the proposed algorithm can match the level of MMSE precoding under perfect channel state information and non-perfect state information.