移动边缘计算中隐私感知的在线任务卸载机制

邓慧娜; 叶阿勇; 刘燕妮; 孙明辉

引用本文：

邓慧娜,叶阿勇,刘燕妮,孙明辉.移动边缘计算中隐私感知的在线任务卸载机制[J].信息安全学报,2023,8(4):126-138 [点击复制]
DENG Huina,YE Ayong,LIU Yanni,SUN Minghui.Privacy-aware online task offloading mechanism in mobile edge computing[J].Journal of Cyber Security,2023,8(4):126-138 [点击复制]

本文已被：浏览 2431次下载 2249次	码上扫一扫！
移动边缘计算中隐私感知的在线任务卸载机制
邓慧娜^1,2, 叶阿勇^1,2, 刘燕妮^1,2, 孙明辉^1,2
0 字体:加大+\|默认\|缩小-
(1.福建师范大学计算机与网络空间安全学院福州中国 350117;2.福建省网络安全与密码技术重点实验室福州中国 350117)

摘要:

移动边缘计算是一种新型的计算范式,它将云计算能力从集中式云分布到网络边缘,可有效解决云计算实时性低及移动终端计算能力不足等问题。但由于用户移动的不确定性以及边缘服务器的覆盖范围的有限性,使得实现高效率的任务卸载面临挑战,并且现有可用性优先的任务卸载算法容易造成用户轨迹隐私泄露。针对上述问题,本文考虑了迁移成本、轨迹隐私与可用性三者之间的矛盾关系,基于信息论提出一种高可用性的在线隐私感知任务卸载机制。首先,基于真实轨迹与发布轨迹之间的互信息量化轨迹隐私泄露程度,并将该任务卸载问题转换为多目标优化问题;然后,进一步提出一种基于马尔可夫链的任务卸载方案来求解该优化问题;最后,在多约束场景下设计了面向设备端的轻量级在线任务卸载算法,解决了在迁移成本约束下轨迹隐私与感知时延的加权平衡问题,以及迁移成本与感知时延双重约束下的轨迹隐私泄露最小化问题。实验结果表明,本文提出的隐私感知任务卸载方案在不同约束场景下的安全性均优于其他方案,能以较低的感知时延实现轨迹隐私保护,适用于资源受限的移动设备进行快速决策与卸载。

关键词: 移动边缘计算|轨迹隐私|迁移成本|可用性|任务卸载

DOI：10.19363/J.cnki.cn10-1380/tn.2023.07.09

投稿时间：2022-01-07修订日期：2022-02-21

基金项目:本课题得到资助国家自然科学基金(No. 61972096, No. 61771140, No. 61872088, No. 61872090), 福建省科技厅高校产学合作计划项目(No. 2022H6025)。

Privacy-aware online task offloading mechanism in mobile edge computing

DENG Huina^1,2, YE Ayong^1,2, LIU Yanni^1,2, SUN Minghui^1,2

(1.College of computer and Cyber Security, Fujian Normal University, Fuzhou, China, 350117;2.Fujian Provincial Key Laboratory of Network Security and Cryptology, Fuzhou, China, 350117)

Abstract:

Mobile edge computing is a new computing paradigm, it pushes the computing power of cloud servers from the upper-layer centralized cloud to the lower-layer network edge, which can effectively solve the problems of low real-time performance of cloud computing and insufficient computing power of mobile devices. But due to the uncertainty of the user’s movement and the limited signal coverage of the edge server, it has become a very challenging problem to get the task offload to be completed efficiently. At the same time, there are currently some usability-first task offloading algorithms, which ignore user trajectory privacy and security in order to obtain the lowest delay. In view of these challenges, we considered the contradictory relationship among the three aspects of migration cost, trajectory privacy, and usability. And we propose a privacy-aware online task offloading mechanism with high availability on the basis of information theory. First of all, we quantify the degree of leakage of trajectory privacy by calculating the mutual information between the real trajectory and the released trajectory, and formulate this task offloading problem as a multi-objective optimization problem. Next, we further propose a task offloading scheme based on Markov chain to solve this optimization problem. At last, we designed a lightweight online task offloading algorithm for the device side in a multi-constraint scenario. We solve the weighted balance problem between trajectory privacy and perceived delay under the constraints of migration cost, and the problem of minimizing trajectory privacy leakage under the dual constraints of migration cost and perceived delay. Our experimental results show that, the privacy-aware task offloading scheme proposed in this paper is more secure than other schemes in different constrained scenarios. It can realize trajectory privacy protection while ensuring low perception delay, and it is suitable for fast decision-making and offloading on those mobile devices with limited resources.

Key words: mobile edge computing|track privacy|migration cost|usability|task offloading