Joint compressed-sensing and matrix-completion for efficient data collection in WSNs

Wireless sensor networks have gained considerable interest in the last few years, serving a large number of applications. Data collection efficiency is of paramount importance as sensors are severe resource-constrained devices, and current protocol inefficiencies lead to significant packet loss. In this work, we minimize the necessary information sensors transmit by applying the compressed sensing principles. Moreover, missing information due to packet loss is efficiently recovered using the matrix completion theory. The performance evaluation shows that when these advanced signal processing techniques are jointly used, the reconstruction error is small for high compression ratios, and fairly high packet loss. At the same time, the total energy consumption of the network substantially decreases.