Privacy-preserving quality prediction for edge-based IoT services

Quality of Service (QoS) prediction and privacy preservation are two key challenges in service recommendation. Nevertheless, the existing QoS prediction methods cannot be directly utilized in edge computing networks (ECNs) due to the user mobility and distribution nature in these networks. To address this problem, the DEQP2 model, a distributed edge QoS prediction model with privacy-preserving, is proposed in this paper. In the DEQP2 model, the Laplace vector mechanism is first employed for distributed privacy protection processing at the edge. Then, the distributed edge differential privacy (DEDP) QoS prediction algorithm is proposed, to enable a comprehensive consideration of the influence of user preferences and the edge environment. Finally, we conduct experiments on the EdgeQoS dataset, which is an integrated dataset derived from the WSDream and Telecom real-world datasets. The experimental results show that the proposed DEQP2 model provides measurable privacy preservation without significantly reducing the accuracy. (C) 2020 Elsevier B.V. All rights reserved.

Automated summary

The DEQP2 model is a distributed edge QoS prediction model with privacy protection, employing the Laplace vector mechanism and distributed edge differential privacy algorithm. Experiments conducted on the EdgeQoS dataset showed that the DEQP2 model provides measurable privacy preservation without significantly reducing accuracy.