孙鉴, 武涛, 吴隹伟,等.基于改进猫群优化算法的多目标边缘任务调度研究[J].河南理工大学学报（自然科学版）,2025,44(4):29-39.

SUN J, WU T, WU Z W, et al. Research on multi-objective edge task scheduling based on an improved cat swarm optimization algorithm [J]. Journal of Henan Polytechnic University (Natural Science) , 2025, 44(4): 29-39.

基于改进猫群优化算法的多目标边缘任务调度研究

孙鉴1,2, 武涛1, 吴隹伟1, 杨晓焕1, 马宝全1

1.北方民族大学 计算机科学与工程学院，宁夏 银川  750021；2.北方民族大学 图像图形智能处理国家民委重点实验室，宁夏 银川 750021

摘要: 目的 为了在资源有限的边缘计算服务中降低任务传输时延并提高边缘端资源利用率，   方法 提出一种基于改进猫群优化算法（improved cat swarm optimization，ICSO）的边缘计算多目标任务调度方法。在边缘计算任务调度模型上，采用ICSO对其编码求解，引入非线性选择法更新猫群行为比例分配和记忆池，平衡算法的全局搜索能力和局部搜索能力；为弥补算法初始解空间过小的缺陷，引入反向学习策略和扩大算法寻优空间；通过基于平均适应度值的机制改进跟踪行为，增强全局寻优能力，防止陷入局部最优；提出新的被收养行为，使猫群个体空间发生变异和扩散，进一步提高算法寻优能力。  结果 仿真实验结果表明，与现有任务调度算法PPCSO，OBL_TP_PSO，PCSO，DMOOTC，LCSO和CSO相比，ICSO的任务传输时延分别降低了4.3%，7.8%，8.3%，9.3%，10.8%，12.5%，减少了任务的最大完工时间和成本花费，同时可在有限迭代次数内收敛到比其他算法更优的解，验证了算法的可行性。   结论 在应对边缘计算环境中的任务调度问题时，提出的算法优化策略具有一定效果，改进后的猫群优化算法能够提高任务传输效率，使边缘端资源得到更充分利用。 

关键词:边缘计算;任务调度;猫群优化算法;任务传输时延;多目标

doi: 10.16186/j.cnki.1673-9787.2024070045

基金项目:国家自然科学基金资助项目（62062002）；宁夏自然科学基金资助项目（2022AAC03289）；北方民族大学研究生创新项目（YCX23165）

收稿日期:2024/07/10

修回日期:2024/09/23

出版日期:2025/06/19

Research on multi-objective edge task scheduling based on an improved cat swarm optimization algorithm

Sun Jian1,2, Wu Tao1, Wu Zhuiwei1, Yang Xiaohuan1, Ma Baoquan1

1.School of Computer Science and Engineering， North Minzu University， Yinchuan  750021， Ningxia， China； 2.The Key Laboratory of Images and Graphics Intelligent Processing of State Ethnic Affairs Commission， Yinchuan  750021， Ningxia， China

Abstract: Objectives This study aims to reduce task transmission latency and improve resource utilization in resource-constrained edge computing environments. Methods A multi-objective task scheduling method based on an Improved Cat Swarm Optimization （ICSO） algorithm is proposed. The ICSO algorithm is applied to encode and solve the edge task scheduling model. the improved Cat Swarm Optimization algorithm was used for encoding and solving. A nonlinear selection strategy is introduced to update the proportion of cat behaviors and the memory pool， thereby balancing global and local search capabilities. To overcome the limitation of a narrow initial solution space， a reverse learning mechanism is incorporated to expand the optimization search space. In addition， the tracking behavior is enhanced based on average fitness， improving global search ability and avoiding local optima. A novel adoption behavior is also proposed to promote mutation and diffusion among cat individuals， further enhancing the optimization capability. Results Simulation experiments show that， compared with existing task scheduling algorithms such as PPCSO， OBL_TP_PSO， PCSO， DMOOTC， LCSO， and CSO， the proposed ICSO reduces task transmission latency by 4.3%， 7.8%， 8.3%， 9.3%， 10.8%， and 12.5%， respectively. It also reduces the maximum task completion time and cost， and achieves better convergence within a limited number of iterations， demonstrating the effectiveness and feasibility of the approach.  Conclusions The proposed optimization strategy proves effective for task scheduling in edge computing scenarios. The improved Cat Swarm Optimization algorithm significantly enhances task transmission efficiency and ensures more efficient utilization of edge resources. 

Key words: edge computing; task scheduling; cat swarm optimization algorithm; task transmission latency; multi-objective