K-anonymity privacy-preserving algorithm for IoT applications in virtualization and edge computing

Efficient privacy-preserving algorithms are used in internet of things (IoT) applications, virtualization and edge computing environments to decrease the high data disclosure risk. However, due to the strict security principles of current privacy-preserving algorithms, they ignore the time consumption. On the other hand, most existing privacy-preserving algorithms suffer from the overgeneralization problem, creating unnecessary information loss. Thus, two privacy-preserving algorithms are proposed in this paper for IoT applications in the virtualization and edge computing environment to address this problem and balance the information loss and disclosure risk. We first propose an overall metric to measure the performance of privacy-preserving algorithms and balance the information loss and disclosure risk. Second, the proposed algorithms can significantly decrease the time consumption by utilizing cached anonymized datasets. Moreover, the proposed algorithms also optimize the distribution of anonymized datasets and decrease the information loss by deleting the small equivalent classes. Experimental results show that our proposed algorithms are relatively fast, only consuming 34.3% operation time of current algorithms. Next, we tested the usability of anonymized datasets by putting them into an IoT application. The two datasets generate similar results, indicating that the proposed algorithms can satisfy the information loss requirements of IoT applications.