Achieving Efficient and Privacy-Preserving Max Aggregation Query for Time-Series Data
Yunguo Guan, Rongxing Lu, Yandong Zheng, Jun Shao, Guiyi Wei · 2020
The vision of future intelligent information society will be globally data driven, enabled by Internet of Things (IoT) techniques. In any IoT-enabled applications, huge volumes of time-series data are continuously generated by IoT devices, which will be fed for high-level functions. Among these functions, the max aggregation query over a specific time interval is one of the frequently used operations. However, due to the limited resources in IoT devices, a common way to deal with the max aggregation query is to involve powerful cloud servers. Nevertheless, as the sensed data from IoT devices and its pattern (e.g., local ranking sequences) are usually private and the cloud servers are not fully trusted, the data should be encrypted before being outsourced to cloud servers. Obviously, the data encryption will incur some efficiency issues. In this paper, to mitigate the privacy and efficiency issues, we propose an efficient and privacy-preserving max aggregation query scheme for time-series data in IoT scenarios. Specifically, we first employ a segment tree based data structure to represent the data collected by IoT devices. Then, to protect the privacy, we leverage two encryption techniques to encrypt the data structure. With the encrypted data structure, our proposed scheme can handle a ranged max aggregation query with O(log L) time complexity, where L is the range length of the query. Detailed security analysis and performance evaluation show that our scheme can not only preserve the privacy of data and its local ranking sequences, but also achieve efficient ranged max aggregation query.