【作者】 赵辉；

【导师】 张德干； 赵洪祥；

【作者基本信息】 天津理工大学 ， 计算机技术（专业学位）， 2025， 硕士

【摘要】 随着物联网(Internet of Things,IoT)技术近年来的快速发展,涌现出许多智能服务和应用,例如增强现实、虚拟现实、智能制造、自动驾驶车辆、移动医疗、智能家居和智慧城市。这些服务和应用经常包含大量计算密集型和资源密集型任务,需要强大的处理和存储能力。在此背景下,移动计算应运而生,成为支持这些智能应用的重要技术。移动计算通过充分利用边缘和云端的计算资源,显著降低计算延时并提升用户体验。本文研究了物联网中移动边缘计算策略的任务处理方法,采用完全卸载策略来优化时延和能耗,取得了一定成果。论文的主要贡献和创新点如下:(1)针对智慧城市中的物联网应用场景,本文在第三章提出一种基于改进鹈鹕优化算法(Improved Pelican Optimization Algorithm,IPOA)的计算卸载方法。IPOA通过引入0)9)混沌映射生成初始个体,并结合0)飞行机制,增强算法的全局搜索功能和局部开发功能,进而实现更有效的计算卸载策略。该算法能够在复杂网络环境下,快速地找到计算任务的卸载方案,有效平衡任务处理时间和资源利用率。实验结果表明,IPOA在降低时延和能耗方面相较于传统优化算法有显著的优势,特别是在多任务复杂场景中表现出更优异的性能。(2)针对物联网中的云边协同计算卸载场景,本文在第四章提出一种基于蛇鹫优化算法(Secretary Bird Optimization Algorithm,SBOA)和差分进化算法(Differential Evolution,DE)的联合计算卸载方法,即蛇鹫-差分联合优化算法。该算法的第一阶段利用蛇鹫优化算法进行初步解的生成,以快速找到具有较高适应度的解;第二阶段使用差分进化算法对初步解进行进一步优化,以提高解的精度,从而提升系统资源的利用率。该联合方法结合了两种算法的优点,既具备全局搜索的能力,又能在局部开发阶段找到更优的解,为物联网云边协同计算卸载问题提供了高效的解决方案。实验结果表明,蛇鹫-差分联合优化算法在云边协同场景中的性能优越,能够有效降低任务卸载的系统时延和能耗,提高系统整体的资源利用效率。本文的研究为物联网环境下的计算卸载问题提供了新的解决思路,对提高智慧城市等应用场景中的服务响应速度和降低网络资源消耗具有重要的参考意义。更多还原

【Abstract】 In recent years,the rapid evolution of IoT(Internet of Things)technology has led to the rise of various intelligent solutions and innovative applications,including augmented and virtual reality,advanced manufacturing,autonomous driving,telehealth,smart living spaces,and urban intelligence.These services and applications often involve large numbers of computationally and resource-intensive tasks,requiring considerable processing and storage capabilities.In this scenario,mobile edge solutions have become essential for enabling advanced services.They help decrease processing delays and improve user interactions by efficiently leveraging both edge and cloud infrastructures.Mobile computing can significantly reduce computation latency and enhance user experience by effectively utilizing both edge and cloud computing resources.This thesis investigates the methods for IoT computation task offloading based on mobile edge computing strategies.It adopts a full offloading strategy to comprehensively optimize latency and energy consumption.The key contributions and innovations of this research are summarized as follows:(1)For IoT applications in the smart city context,this thesis presents,in Chapter3,a computation offloading method based on the Improved Pelican Optimization Algorithm(IPOA).By introducing the tent chaotic map to generate the initial population and combining it with the levy flight strategy,IPOA enhances the algorithm’s global search ability and local exploitation capability,thereby achieving a more efficient computation offloading strategy.This algorithm can quickly determine offloading schemes for computation tasks in complex network environments,effectively balancing task processing time and resource utilization.Experimental results show that IPOA has significant advantages over traditional optimization algorithms in reducing latency and energy consumption,particularly demonstrating superior performance in multi-task complex scenarios.(2)In Chapter 4,a computation offloading approach for collaborative cloud-edge environments in IoT is introduced,combining the Secretary Bird Optimization(SBOA)with Differential Evolution(DE),collectively named the SBO-DE Algorithm.In the first stage,SBOA is employed to identify promising initial candidates,aiming for high fitness values.In the second phase,DE is utilized to further optimize the initial solutions,thereby improving the solution’s accuracy and optimizing system resource utilization.This combined approach integrates the advantages of both algorithms,providing both strong global search capabilities and effective local exploitation to find better solutions.It offers an efficient solution to the collaborative cloud-edge computing offloading problem in IoT.Experimental results indicate that the Secretary Bird-Differential Evolution Optimization Algorithm performs excellently in collaborative cloud-edge scenarios,effectively reducing system latency and energy consumption while improving overall resource utilization efficiency.This research provides a new approach to the computation offloading problem in IoT environments,offering significant reference value for enhancing service response speed and reducing network resource consumption in applications such as smart cities.更多还原