Mobile Sensing of Two-Dimensional Bandlimited Fields on Random Paths

Mobile sensing has been recently proposed for sampling spatial fields, where mobile sensors record the field along various paths for reconstruction. Classical and contemporary sampling typically assumes that the sampling locations are approximately known. This work explores multiple sampling strategies along random paths to sample and reconstruct a two dimensional bandlimited field. Extensive simulations are carried out, with insights from sensing matrices and their properties, to evaluate the sampling strategies. Their performance is measured by evaluating the stability of field reconstruction from field samples. The effect of location unawareness on some sampling strategies is also evaluated by simulations.
Comment: 6 pages, 2 figures; submitted to ICASSP 2018