基于ADMM的低轨多卫星系统分布式预编码设计

ADMM-Based Distributed Precoding Design for LEO Multi-satellite Systems

  • 摘要: 针对具有星间链路(inter-satellite links,ISL)的低轨(low earth orbit,LEO)多卫星系统,提出了一种基于多卫星协作传输的和速率(sum rate,SR)最大化预编码算法. 传统的预编码算法需要复杂的星上计算来得到数值解,这导致低轨卫星系统面临较大的计算开销和延迟问题. 为解决上述关键问题,设计了一种基于交替方向乘子法(alternating direction method of multipliers,ADMM)的高吞吐量、低复杂度、具有闭式解的分布式预编码算法. 该算法通过构建辅助变量和问题分解,将预编码设计问题转化为多个子问题并行求解,每个子问题仅有一个约束条件,并在每次迭代后仅通过星间链路交换设计的数据矩阵,从而有效实现分布式预编码. 仿真结果表明,与典型的两步和速率最大化算法相比,所提出的算法可以实现更高的和速率,同时大幅降低计算复杂度.

     

    Abstract: A distributed pre-coding algorithm was proposed for low earth orbit (LEO) multi-satellite systems with inter-satellite links (ISLs). Due to the complicated on-board calculations for the numerical solutions in traditional pre-coding algorithms and induced unacceptable computational cost and delay in LEO satellite systems, a distributed pre-coding design algorithm was proposed based on the alternating direction method of multipliers (ADMM) to achieve low complexity, high-throughput and the closed-form solutions. In detail, constructing the auxiliary variables and decomposing the problem, the algorithm was arranged to transform the pre-coding design problem into multiple sub-problems and solve them synchronously. And then, providing one constraint item for each sub-problem and exchanging the designed data matrices only via ISLs after each iteration, the algorithm was designed to carry out distributed pre-coding effectively. Simulation results show that the proposed algorithm can achieve a higher beam rate with lower complexity than that achieved with the typical two-step SR maximization algorithm.

     

/

返回文章
返回