Exploring Fundamental Limits of Spatiotemporal Sensing for Non-Linear Inverse problems.

This paper considers the problem of identifying the joint support set (of size K) of N unknown sparse vectors with d entries from 1-bit M compressed spatial measurements and N temporal measurements. We propose an adaptive method which can recover the joint support exactly provided that total spatiotemporal measurements satisfy MN = K log(d). Unlike most 1-bit MMV algorithms, our strategy can operate in the extreme spatial compression regime M 1 .