| Visible Light Communication(VLC)is one of the complementary technologies of indoor network,with rich spectrum resources,strong electromagnetic interference resistance and other advantages.For indoor VLC multi-user MIMO system,there are many problems such as multi-user interference(MUI)and Block Diagonalization(BD)algorithms with high computational complexity.This paper studies the improved BD algorithm with low complexity.The specific research work is as follows:1.Aiming at the problem that BD algorithm is too complicated to perform SVD twice for each user,this paper summarizes the existing low-complexity improved BD algorithms,and on the basis of the existing improved algorithms,this paper proposes an improved BD algorithm based on the combination of QR decomposition and LDL decomposition of Gram-Schmidt Orthogonalization(GSO-LDL-BD).Through complexity analysis and simulation verification,compared with the traditional BD algorithm,the complexity of GSO-LDL-BD algorithm is reduced by 43.87%without loss of capacity.2.Aiming at the problem that GSO-LDL-BD algorithm has no optimization in capacity and bit error rate.In this paper,LLL lattice reduction algorithm based on Minimum Mean Square Error(MMSE)is introduced to improve GSO-LDL-BD algorithm,and two improved BD algorithms are proposed.An improved BD algorithm,which combines QR decomposition based on GSO orthogonalization with LLL lattice reduction algorithm based on MMSE criterion(GSO-LLL-BD),can obtain higher performance gain because the matrix has better orthogonality after lattice reduction operation.The results show that the complexity of GSO-LDL-BD algorithm is further reduced by 6.86%,the system and capacity are increased by 5-15bit/s/Hz,and the bit error rate is reduced by an order of magnitude.The second improved BD algorithm,QR decomposition realized by Givens transformation is introduced into BD algorithm,and an improved BD algorithm is proposed,which combines QR decomposition based on Givens decomposition and LLL lattice protocol operation based on MMSE criterion(Givens-LLL-BD).Through matrix decomposition,QR decomposition based on Givens has better orthogonality than orthogonal matrix obtained by GSO orthogonal decomposition,further improving system and capacity and bit error rate performance.Through complexity analysis and simulation verification,compared with GSO-LLL-BD algorithm,although the complexity of Givens-LLL-BD algorithm increases by 5.24%,the system capacity increases by 3-10 bit/s/Hz and the bit error rate performance decreases by an order of magnitude. |
