基于自适应算子与非支配排序的星地网络多目标优化方法  

Adaptive Operator Adjustment and Non-dominated Sorting Based Multi-objective Optimization for Satellite-Terrestrial Systems

作　　者：李浩辉 周雨 田金今 赵晓辉[1] 雷磊[1,3] LI Haohui;ZHOU Yu;TIAN Jinjin;ZHAO Xiaohui;LEI Lei(School of Information and Communications Engineering,Xi'an Jiaotong University,Xi'an 710049,China;Department of Computing,The Hong Kong Polytechnic University,Hong Kong 999077,China;The National Mobile Communications Research Laboratory,Southeast University,Nanjing 210096,China)

机构地区：[1]西安交通大学信息与通信工程学院,陕西西安710049 [2]香港理工大学电子计算学系,中国香港999077 [3]东南大学移动通信国家重点实验室,江苏南京210096

出　　处：《无线电通信技术》2023年第5期809-815,共7页Radio Communications Technology

基　　金：东南大学移动通信国家重点实验室开放研究基金(2023D02);四川省自然科学基金(2023NSFSC0455)。

摘　　要：低轨(Low Earth Orbit,LEO)高通量卫星将成为未来非地面网络的重要组成部分。基于星上灵活载荷技术,探索未来6G星地网络动态高效的多维资源调度和多指标优化方法。实际优化过程中,多个指标间如吞吐量、接入用户数等往往相互冲突。已有工作多通过加权求和将多个优化目标转换为单目标优化问题进行求解。然而此种方法通常存在不同目标函数之间量纲单位无法统一,病态矩阵造成各目标函数权重无法准确分配,以及在大规模优化中难以接近帕累托前沿等问题。针对星地融合网络如何同时提高终端接入数量与提高多用户吞吐的优化问题,建立单目标和多目标优化模型,通过对单目标优化问题分配不同权重,进行最优求解,产生多组优化结果作为性能比较的基准方案之一,提出基于快速非支配排序与自适应算子调整的高效多目标优化算法。实验结果表明,所提算法较加权单目标优化与传统多目标优化算法,可有效提升整体的优化性能,进而提升星地融合网络的通信性能。As a consensus of 6G,Low Earth Orbit(LEO) satellites will become an important component of non-terrestrial networks.Based on advances of flexible satellite payload,this paper explores dynamic and efficient multi-dimensional resource scheduling and multi-objective optimization approaches for future 6G satellite-terrestrial networks.In practice,operators need to optimize multiple metrics simultaneously,such as throughput and the number of served users.These metrics typically conflict with each other.Most previous works transform multiple optimization objectives into a single objective through weighted summation,e.g.,solving a single-objective optimization problem.However,this approach encounters several issues,such as non-unified units among objective functions,inaccurate assignment of weights to different objectives,and difficulties in approaching the Pareto front in large-scale optimization.In this paper,we first formulate mathematical models based on single-objective and multi-objective optimization.We optimally solve the single-objective optimization problem and generate results with different weight assignments as performance benchmarks.Subsequently,we propose a non-dominated sorting and adaptive operator adjustment based algorithm to simultaneously solve multiple objective functions and compare the results with typical conventional approaches.

关 键 词：低轨卫星 资源调度 组合优化 多目标优化 

分 类 号：TN919.23[电子电信—通信与信息系统]