Energy-Aware Sampling Set Selection for Signal Reconstruction in Internet of Things

Emerging technologies surge the rapid development of the Internet of things (IoT). In this paper, we apply graph signal processing (GSP) to study the sampling set selection problem in an IoT network where signals are collected from massive sensor nodes and reconstructed at an aggregation center. Based on the eigen-decomposition-free sampling method, the Chebyshev polynomial approximation for signal reconstruction is employed to develop a sampling set selection algorithm for a sensor network with energy-constrained sensor nodes. In the proposed algorithm, sensors are selected not only according to the short-term signal reconstruction accuracy, but also determined by the energy remained in each sensor. Specifically, in the formulated sampling-set-selection optimization problem, the energy-aware penalty weights for IoT sensors are designed to trade off the reconstruction accuracy and energy consumption. Simulation results show that the proposed algorithm outperforms random selection and the conventional sampling set selection method.