Heterogeneous networked data recovery from compressive measurements using a copula prior
This publication appears in: IEEE Transactions on Communications
Authors: N. Deligiannis, J. Mota, E. Zimos and M. Rodrigues
Publication Date: Dec. 2017
Large-scale data collection by means of wirelesssensor network and internet-of-things technology poses variouschallenges in view of the limitations in transmission, computation,and energy resources of the associated wireless devices. Compressivedata gathering based on compressed sensing has beenproven a well-suited solution to the problem. Existing designsexploit the spatiotemporal correlations among data collected bya specific sensing modality. However, many applications, suchas environmental monitoring, involve collecting heterogeneousdata that are intrinsically correlated. In this study, we proposeto leverage the correlation from multiple heterogeneous signalswhen recovering the data from compressive measurements.To this end, we propose a novel recovery algorithmbuiltupon belief-propagation principlesthat leverages correlatedinformation from multiple heterogeneous signals. To efficientlycapture the statistical dependencies among diverse sensor data,the proposed algorithm uses the statistical model of copulafunctions. Experiments with heterogeneous air-pollution sensormeasurements show that the proposed design provides significantperformance improvements against state-of-the-art compressivedata gathering and recovery schemes that use classical compressedsensing, compressed sensing with side information, anddistributed compressed sensing.