CBRN Data Fusion Using Puff-Based Model and Bar-Reading Sensor Data

Authors

Cheng, Y., Reddy, K. V. U., Singh, T. & Scott, P.

Source

10th International Conference on Information Fusion

Abstract

This article provides an approximate approach to the measurement update of the puff state in the data assimilation process for airborne material dispersion where the sensor measurements are bar readings. Based on the Bayes rule and numerical quadrature techniques, the approach approximates an interval in the concentration space associated with a sensor’s bar reading with a set of discrete points and the integrals over the interval by sums of function evaluations at those points. An alternative approximation to the integrals involves the Gaussian error function and the Hermite-Gaussian quadrature. Under the assumption that all the distributions are approximately Gaussian, a two-step procedure of the update is presented: 1) updating the continuous-valued concentration forecast with the bar-reading data and 2) updating the puff state based on the correlation between the puff state and the concentration.