The stability of Kalman filters under randomly lost measurements has been studied extensively, but most work has focused on discrete-time or specialized loss models. This paper revisits the problem in a general continuous-time framework, where both the measurement matrix and the noise covariance are allowed to be random. That broader setup captures more realistic sensing scenarios, such as randomized sensor scheduling or intermittent connectivity.

By treating the measurement process as randomized, the authors aim to identify fundamental limits on filter stability and to translate those limits into practical guidance for sensor network design. The abstract indicates that the work targets field estimation, where the spatial distribution of sensors and the random availability of their measurements directly affect how well a dynamic field can be tracked.

Because the abstract is truncated, the specific theorems and conditions are not available in the source. Still, the framing suggests a unified treatment that could help engineers choose sensor layouts and sampling strategies that guarantee stable estimation even when measurements are unreliable.