Sparse modeling and optimization tools for energy efficient and reliable IoT

The continuous expansion of the Internet of Things, with even more devices expected to connect in the following years, leads to a great increase in the amount of data required to be transmitted and stored, identifying the need for robust, reliable and energy efficient data flow in the context of an ever-growing network. This study presents a matrix completion based approach for big data and large matrices facilitating local smoothness constraints combined with active subspace computing. We demonstrate that the presented approaches exhibit lower reconstruction error with respect to relevant literature for the specific problem. We also examine the effect of matrix size on the reconstruction accuracy so as to investigate the suitability of the proposed approaches for big data.