Multi-Attribute Data Recovery in Wireless Sensor Networks With Joint Sparsity and Low-Rank Constraints Based on Tensor Completion

Jingfei He,Tianyu Geng,Guiling Sun,Yatong Zhou

doi:10.1109/access.2019.2942195

Abstract

In wireless sensor networks (WSNs), data recovery is an indispensable operation for data loss or energy constrained WSNs using sparse sampling. However, the recovery accuracy is not satisfying for WSNs with various sensor types due to the neglect of the correlation among multi-attribute data. In this paper, we propose a novel data recovery method with joint sparsity and low-rank constraints based on tensor completion for multi-attribute data in WSNs. The proposed method represents the high-dimensional data as low-rank tensors to effectively exploit the correlation that exists in the multi-attribute data. The utilization of the spatiotemporal sparsity in the signal is emphasized by sparsity constraints. Furthermore, an algorithm based on the alternating direction method of multipliers is developed to solve the resultant optimization problem efficiently. Experimental results demonstrate that the proposed method significantly outperforms existing solutions in terms of recovery accuracy in WSNs.

Highlights

Wireless sensor networks (WSNs) have been widely used in numerous applications including military surveillance, environmental monitoring, and health care monitoring, in which a number of sensor nodes monitor physical phenomena and transmit the data to a base station or sink node for processing [1], [2]
EXPERIMENTS AND ANALYSIS In order to evaluate the effectiveness of the proposed method for recovery of multi-attribute data in WSNs, the matricization methodologies including the sparsity constraint method [5] and the matrix completion method [8] and the tensor methodologies including the smooth PARAFAC tensor completion (SPC) [34], the low-rank tensor completion (LRTC) [43] and tensor Singular Value Decomposition (t-singular value decomposition (SVD)) [23], [27] are chosen for comparisons
In this paper, we propose a novel data recovery method with joint sparsity and low-rank constraints based on tensor completion to increase the recovery accuracy for multi-attribute data in WSNs

Summary

Introduction

Wireless sensor networks (WSNs) have been widely used in numerous applications including military surveillance, environmental monitoring, and health care monitoring, in which a number of sensor nodes monitor physical phenomena and transmit the data to a base station or sink node for processing [1], [2]. Due to the hardware and wireless conditions, data loss is common in WSNs, especially in large scale WSNs. Besides, to further reduce the energy consumption in energy constrained WSNs, one straightforward way is using sparse sampling to reduce the number of measurement (i.e., the collected data) [3], [4]. To further reduce the energy consumption in energy constrained WSNs, one straightforward way is using sparse sampling to reduce the number of measurement (i.e., the collected data) [3], [4] Both data loss situation and utilization of sparse sampling method result in the recovery problem to estimate the missing data in WSNs. as an indispensable and important operation, data recovery becomes one of the key research issues in WSNs. The associate editor coordinating the review of this manuscript and approving it for publication was Xianfu Lei

Methods

Findings

Discussion

Conclusion