Resource allocation for multi-variate dynamic Gaussian estimation
David Lucking, Nathan A. Goodman · 2018
Noisy data require estimation techniques to infer knowledge about a parameter's value. Furthermore, many sensors encounter environments with multiple dynamic variables, which adds additional complexity in the form of timeline and resource allocation constraints. When these multiple parameters require distinct illumination, the system must choose the power allocated for each measurement, subject to a total power constraint, to prioritize parameters that are important or desperately in need of an update. By carefully allocating the power between multiple measurements, this paper details strategies to minimize the overall uncertainty under different conditions of SNR, available timeline, and strength of the parameters' dynamics. We also consider allocation strategies that attempt to maintain all parameters below an uncertainty threshold.