An Algorithm for Sensor Data Uncertainty Quantification

This letter presents an algorithm for reducing measurement uncertainty of one physical quantity when oversampling measurements two quantities with correlated noise. The assumes that the aleatoric in both follows a Gaussian distribution and relies on sampling faster than it is possible measurand (the true value we are trying to measure) change (due system thermal time constant) calculate parameters noise distribution. In contrast Kalman particle filters, which, respectively, require state update equations map quality, our requires only oversampled sensor measurements. When applied temperature-compensated humidity sensors, provides reduced estimates from temperature experimental evaluation, achieves average reduction 10.3%. incurs execution overhead 5.3% compared minimum required measure uncertainty. Detailed instruction-level emulation C-language implementation compiled RISC-V architecture shows program 0.05% more instructions per iteration operations