当前位置: X-MOL 学术Eng. Appl. Artif. Intell. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Towards privacy preserving AI based composition framework in edge networks using fully homomorphic encryption
Engineering Applications of Artificial Intelligence ( IF 7.5 ) Pub Date : 2020-06-24 , DOI: 10.1016/j.engappai.2020.103737
Mohammad Saidur Rahman , Ibrahim Khalil , Mohammed Atiquzzaman , Xun Yi

We present a privacy-preserving framework for Artificial Intelligence (AI) enabled composition for the edge networks. Edge computing is a very promising technology for provisioning realtime AI services due to low response time and network bandwidth requirements. Due to the lack of computational capabilities, an edge device alone cannot provide the complex AI services. Complex AI tasks should be divided into multiple sub-tasks and distributed among multiple edge devices for efficient service provisioning in the edge network. AI-enabled or automatic service composition is one of the essential AI tasks in the service provisioning. In edge computing-based service provisioning, service composition related tasks need to be offloaded to several edge nodes for efficient service. Edge nodes can be used for monitoring services, storing Quality-of-Service (QoS) data, and composing services to find the best composite service. Existing service composition methods use plaintext QoS data. Hence, attackers may compromise edge devices to reveal QoS data of services and modify them for giving an advantage to particular edge service providers, and the AI-based service composition becomes biased. From that point of view, a privacy-preserving framework for AI-based service composition is required for the edge networks. In our proposed framework, we introduce an AI-based composition model for edge services in the edge networks. Additionally, we present a privacy-preserving AI service composition framework to perform composition on encrypted QoS data using fully homomorphic encryption (FHE) algorithm. We conduct several experiments to evaluate the performance of our proposed privacy-preserving service composition framework using a synthetic QoS dataset.



中文翻译:

使用完全同态加密实现边缘网络中基于隐私保护的AI组成框架

我们为边缘网络提供了一种用于人工智能(AI)的组合的隐私保护框架。由于响应时间短和网络带宽要求低,边缘计算是一种非常有前途的技术,可用于提供实时AI服务。由于缺乏计算能力,仅边缘设备无法提供复杂的AI服务。复杂的AI任务应分为多个子任务,并分布在多个边缘设备之间,以在边缘网络中高效地进行服务供应。支持AI或自动服务组合是服务供应中必不可少的AI任务之一。在基于边缘计算的服务供应中,需要将与服务组合相关的任务卸载到几个边缘节点以实现高效服务。边缘节点可用于监视服务,存储服务质量(QoS)数据,并组合服务以找到最佳的组合服务。现有的服务组合方法使用纯文本QoS数据。因此,攻击者可能会破坏边缘设备以显示服务的QoS数据,并对其进行修改以使特定的边缘服务提供者受益,并且基于AI的服务组成也变得偏颇。从这个角度来看,边缘网络需要一个用于基于AI的服务组合的隐私保护框架。在我们提出的框架中,我们为边缘网络中的边缘服务引入了基于AI的组合模型。此外,我们提出了一个保护隐私的AI服务组合框架,以使用完全同态加密(FHE)算法对加密的QoS数据执行组合。

更新日期:2020-06-24
down
wechat
bug