Stochastic sensing: Stochastic sensing is a key enabler for the design of self-organizing networks and emerging IoT applications, including crowd-sensing and environmental monitoring. Our main contributions have focused on the development of a framework for design and analysis of wireless sensor networks, enabling multidimensional signal reconstruction from spatiotemporal stochastic sample. Specific contributions include:
- Generalized Sampling Theorems: Generalized the reconstruction of a stationary random process in one dimension, which, for regular sampling, was addressed by Balakrishnan and Lloyd based on Whittaker-Kotelnikov-Shannon sampling theory, while, for irregular sampling, was described by a Levinson’s theorem establishing the condition for perfect reconstruction.
- Signal sampling and reconstruction in presence of uncertainties: Analyzed the sampling and reconstruction of spatiotemporal signals from randomly gathered samples in a multidimensional space. Derived the optimal interpolator as a function of signal properties (signal spectrum and spatial correlation) and sampling properties (inhomogeneous sensor spatial distribution, sample availability, and non-ideal knowledge of sensors positions). Characterized the impact of locations and measurements uncertainties on spatiotemporal signal reconstruction.