摘要
In this paper, we consider the design of linear precoding in Multiple Input Multiple Output （MIMO） Power Line Communication （PLC） systems with finite-alphabet input. First, we derive mutual information of MIMO-PLC systems with impulsive noise. Considering the non-concavity of the objective function and the low-cost requirement of PLC systems, we choose the lower bound of the mutual information as the objective function. Subsequently, we propose a novel approach to design the precoding scheme to reduce computational complexity. Specifically, our work primarily includes the following two contributions： （1） We design the right unitary matrix that is a product of two fixed unitary matrices, which only depends on the modulation mode. Hence, the results can be saved and require less computations. （2） For the power allocation matrix, we first reduce the space of power allocation using constraints of the optimal power allocation policy. Consequently, we propose a non-linear search method to obtain the optimal power allocation in small space. In regards to the computational complexity of the analysis, we conclude that the proposed precoding matrix design scheme has low complexity and is easy to implement. Moreover, the numerical results are proven to demonstrate the performance of the proposed precoding design scheme.
In this paper, we consider the design of linear precoding in Multiple Input Multiple Output （MIMO） Power Line Communication （PLC） systems with finite-alphabet input. First, we derive mutual information of MIMO-PLC systems with impulsive noise. Considering the non-concavity of the objective function and the low-cost requirement of PLC systems, we choose the lower bound of the mutual information as the objective function. Subsequently, we propose a novel approach to design the precoding scheme to reduce computational complexity. Specifically, our work primarily includes the following two contributions： （1） We design the right unitary matrix that is a product of two fixed unitary matrices, which only depends on the modulation mode. Hence, the results can be saved and require less computations. （2） For the power allocation matrix, we first reduce the space of power allocation using constraints of the optimal power allocation policy. Consequently, we propose a non-linear search method to obtain the optimal power allocation in small space. In regards to the computational complexity of the analysis, we conclude that the proposed precoding matrix design scheme has low complexity and is easy to implement. Moreover, the numerical results are proven to demonstrate the performance of the proposed precoding design scheme.
