基于NSGA-Ⅱ的自适应权值物联网服务组合方法  

An adaptive weighting service composition method for Internet of Things based on NSGA-Ⅱ

作　　者：杨雪薇 江凌云[1,2] 李研 YANG Xuewei;JIANG Lingyun;LI Yan(School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Internet of Things Research Institute,Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Unicom Internet of Things Co.,Ltd.,Nanjing 210003,China)

机构地区：[1]南京邮电大学通信与信息工程学院,南京210003 [2]南京邮电大学物联网研究院,南京210003 [3]联通物联网有限责任公司,南京210003

出　　处：《智能计算机与应用》2024年第6期1-10,共10页Intelligent Computer and Applications

基　　金：江苏省重点研发项目(BE2020084-4)。

摘　　要：由于网络服务种类繁多,用户需求复杂多变,因此在处理面向用户需求的多目标服务组合任务时,根据用户需求偏好对多目标QoS属性进行加权处理得到的初始化目标函数,在子目标的权值处理上会产生一定的偏差,这可能使多目标优化算法在求解服务组合问题时得到的解集不准确,出现不满足用户实际需求的情况。所以本文提出了一种面向用户需求的多目标自适应权值服务组合方法。该方法根据迭代过程中子目标均值与目标期待值之间的相对偏差值,动态修正加权系数。通过权值的修正使构建的目标函数更符合用户需求偏好,同时避免子目标陷入局部最优的困境。仿真结果表明,改进后的算法在帕累托前沿上有明显的优势,避免了局部最优,且时间复杂度远低于其他算法。随着迭代过程中权值的修正,目标函数均值也逐步贴近期待值,有效地满足了用户的需求。Due to the wide variety of network services and the complex and changeable user demands,when dealing with the multi-objective service combination task oriented to user demands,the initial objective function obtained by weighting the multiobjective QoS attributes according to user demand preferences will produce certain deviations in the weight processing of sub-object.This may cause the solution set obtained by the multi-objective optimization algorithm to be inaccurate when solving the service composition problem,which may not meet the actual needs of users.Therefore,this paper proposes a multi-objective adaptive weight service composition method for user requirements.According to the relative deviation between the sub-target mean value and the target expected value in the iterative process,the weighted coefficient is dynamically modified.By modifying the weight value,the constructed objective function is more in line with the user's demand preference,and at the same time,the sub-objective is avoided to fall into the dilemma of local optimization.Simulation results show that the improved algorithm has obvious advantages on the Pareto frontier,avoids local optimization,and the time complexity is much lower than other algorithms.With the weight correction in the iterative process,the mean value of the objective function is also gradually close to the expected value,which effectively meets the needs of users.

关 键 词：多目标优化 自适应权值 服务组合 QOS 

分 类 号：TP393.0[自动化与计算机技术—计算机应用技术]