Modified likelihood probabilistic data association filter for tracking systems with delayed and lost measurements

In this note, a new probabilistic data association filter (PDAF) is developed for single object tracking, where observations are encountered with random delays and losses. When data are transmitted to the filter with latency and dropout, the common likelihood function which extracts information about the state of the intended object from the measurements, cannot obtain accurately the relationship between received observations and the object's state. So, the likelihood function of the PDAF is modified here to cope with delayed and lost measurements. The introduced idea can be used in other filtering methods to prepare them for application in networked tracking systems. Simulation results of two-dimensional scenarios are presented to verify the considerable improvement in the performance of the proposed PDAF.