Dynamic Deployment Optimization of Near Space Communication System Using a Novel Estimation of Distribution Algorithm

Zhao Wang,Maoguo Gong
DOI: https://doi.org/10.1016/j.asoc.2019.02.045
IF: 8.7
2019-01-01
Applied Soft Computing
Abstract:A robust deployment of the airship platforms is crucial to the performance of the Near Space Communication System (NSCS) in the dynamic environment. In this paper, a multiobjective NSCS deployment optimization model with multi-phased periodic user distribution is proposed. To optimize this model, we propose a local incremental estimation of distribution algorithm with an asymmetrical domination relationship within the multiobjective evolutionary algorithm based on decomposition framework. The conflict between the selection mechanism and the domination relationship is also analyzed theoretically for the first time. To obtain robust solutions under this conflict, the local distribution information of a certain subproblem within several generations is encompassed into a local incremental distribution model. As a generalized form of the existing domination relationship, an asymmetrical domination relationship (ADR), which treats the current and past objective values differently, is proposed to select robust solutions. The proposed algorithm is also tested on four designed problems compared with another four popular algorithms and proves its superiority. Some important parameters are also investigated in the experiments and some guidelines on tuning these parameters are given as well.
What problem does this paper attempt to address?