Maximum likelihood track correlation for multisensor integration

This paper addresses multisensor track integration (MSI) in a multitarget environment. This problem differs from the standard in that filtered sensor track files provide the measurements for MSI. Consequently, the sensor track file errors, which typically impact the tracking decisions, are correlated in the time variable. To avoid the unreasonable cost of calculating these autocorrelations, the approach here is to incorporate them into the derivation of the track association log-likelihoods. The result is a maximum likelihood track correlation criterion which, under specified conditions, can be evaluated using only the multisensor track filter variables.