Recovering missing values from corrupted spatio-temporal sensory data via robust low-rank tensor completion

Ruan, Wenjie and Xu, Peipei and Sheng, Quan Z. and Falkner, Nickolas J.G. and Li, Xue and Zhang, Wei Emma (2017) Recovering missing values from corrupted spatio-temporal sensory data via robust low-rank tensor completion. In: Database Systems for Advanced Applications. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) . Springer, CHN, pp. 607-622. ISBN 9783319557526

Abstract

With the booming of the Internet of Things, tremendous amount of sensors have been installed in different geographic locations, generating massive sensory data with both time-stamps and geo-tags. Such type of data usually have shown complex spatio-temporal correlation and are easily missing in practice due to communication failure or data corruption. In this paper, we aim to tackle the challenge-how to accurately and efficiently recover the missing values for corrupted spatiotemporal sensory data. Specifically, we first formulate such sensor data as a high-dimensional tensor that can naturally preserve sensors’ both geographical and time information, thus we call spatio-temporal Tensor. Then we model the sensor data recovery as a low-rank robust tensor completion problem by exploiting its latent low-rank structure and sparse noise property. To solve this optimization problem, we design a highly efficient optimization method that combines the alternating direction method of multipliers and accelerated proximal gradient to minimize the tensor’s convex surrogate and noise’s ℓ1-norm. In addition to testing our method by a synthetic dataset, we also use passive RFID (radiofrequency identification) sensors to build a real-world sensor-array testbed, which generates overall 115,200 sensor readings for model evaluation. The experimental results demonstrate the accuracy and robustness of our approach.