With the development of 5G and the explosive growth of massive Internet of Health Things devices, existing computing resource allocation strategies have many problems such as long delay and poor security performance. Therefore, this paper proposes an optimization strategy for computing resource allocation of massive IoHT devices considering privacy protection in cloud edge computing environment. Firstly, a 5G heterogeneous cloud edge computing network is constructed. Besides, according to network status, the computing requirements of devices are allocated to local execution or edge computing. The computing delay, communication and computing resource allocation of edge servers are modeled accordingly. Then, taking the delay and energy consumption of network computing resource allocation as optimization goal, the priority of subtasks is sorted to realize the optimal allocation of computing resources. Finally, a protection model for instant messaging privacy data is designed by considering the risk of large-scale privacy data leakage in IoHT. Terminal devices under the same local area network are connected to edge servers by Socket without cloud server forwarding, which improves the security performance of privacy data. Experiment and demonstrate the performance of our proposed strategy on MATLAB simulation platform. The results show that the increase of edge computing server will affect the CPU proportion. Moreover, compared with other strategies, the number of users, the number of edge computing servers, the computing capacity of devices and the task arrival rate have the least impact on the average delay of proposed strategy, which effectively improves the performance of allocation strategy.

随着5G的发展以及大规模的物联网设备的爆炸性增长，现有的计算资源分配策略存在许多问题，例如延迟时间长和安全性能差。因此，本文提出了一种在云边缘计算环境中考虑隐私保护的大型IoHT设备计算资源分配的优化策略。首先，构建5G异构云边缘计算网络。此外，根据网络状态，将设备的计算需求分配给本地执行或边缘计算。相应地对边缘服务器的计算延迟，通信和计算资源分配进行了建模。然后，以网络计算资源分配的时延和能耗为优化目标，子任务的优先级被排序以实现计算资源的最佳分配。最后，通过考虑IoHT中大规模隐私数据泄漏的风险，设计了用于即时消息传递隐私数据的保护模型。同一局域网下的终端设备通过Socket连接到边缘服务器，无需云服务器转发，提高了隐私数据的安全性能。在MATLAB仿真平台上进行实验并演示我们提出的策略的性能。结果表明，边缘计算服务器的增加会影响CPU的使用比例。此外，与其他策略相比，用户数量，边缘计算服务器数量，设备的计算能力和任务到达率对所提出策略的平均延迟影响最小，