Grid-based probability density matrix for multi-sensor data fusion

Abstract

The multi-sensor multi-target localization and data fusion problem is discussed, and a novel data fusion method called grid-based probability density matrix (GBPDM) is proposed. Dividing the common observe space into numerous of small grids, the measurements containing uncertainties can be represented by their sampled probability density functions. By adding probability density of all measurements taken from one sensor grid by grid, we got the probability density matrix (PDM) of this sensor. Combining PDMs of all sensors together produce a joint PDM. Peaks in the joint PDM can be considered as estimated locations of targets. Theoretic analysis show that the computation cost is proportional to the product of the number of sensors and that of targets, and will not lead to combinatorial explosion. The presented method has a high precision and is suit for parallel processing. Simulation results verify the feasibility and validity of the proposed technique.