采用有少量反相器和开关(Inverter and Switch,IS)组合的预编码架构设计模拟预编码部分,相比传统移相器结构的混合预编码可以有效降低系统功耗.利用此结构的混合预编码在计算收发端最优的编码矩阵时会变成一个求解复杂的离散组合问题.针对多天线多用户的毫米波大规模多输入多输出(Multiple Input Multiple Output,MIMO)场景,提出了一种新的优化方案——SVD_CE,可将2个矩阵的联合优化问题转化为2个独立的组合优化问题,基于改进交叉熵(Cross Entropy,CE)算法分别求解编解码矩阵.仿真结果表明,所提方案与已有解决方案相比不会造成系统性能的损失,在取得相同性能时利用CE算法中所需候选集的数量大幅减少,有效降低了求解的复杂度.
Low Complexity Hybrid Precoding Scheme Based on Inverter and Switch Structure
The precoding architecture with a few inverters and switches(IS)is used to design the analog precoding part,which can effectively reduce the power consumption of the system compared with the traditional phase shifter structure of the hybrid precoding.The hybrid precoding using this structure becomes a complex discrete combination problem when calculating the optimal encoding matrix at the receiving and sending ends.A new optimization scheme SVD_CE is proposed for multi-antenna and multi-user Multiple Input Multiple Output(MIMO)scenarios,which can transform the joint optimization problem of two matrices into two independent combinative optimization problems.Then the codec matrix is solved based on the improved Cross Entropy(CE)algorithm.The simulation results show that compared with the existing solutions,the proposed scheme will not cause the loss of system performance,and when the same performance is achieved,the number of candidate sets required by the CE algorithm in this scheme is greatly reduced,which effectively reduces the complexity of solution.
millimeter wavemassive MIMOhybrid precodingCE algorithmsingular value decomposition
万方数据
维普
