Efficient data association for 3D passive sensors: If i have hundreds of targets and ten sensors (or more)

This paper considers the passive-sensor data association problem based on multi-dimensional assignment (MDA). The S-D algorithm has been shown to be effective for solving the MDA problem. The bottleneck of the S-D algorithm lies in its cost computation, which consumes about 95%-99% of the CPU times. Since the number of costs in the MDA problem increases exponentially with the number of sensors, the S-D algorithm becomes quite inefficient when a large number of sensors are used. We propose an efficient data association technique, “S₀-D+Seq(2-D)” algorithm, which decomposes the original problem to an S₀-dimensional assignment and several 2-dimensional assignments. The S₀-D+Seq(2-D) algorithm yields a total number of costs which only increases quadratically with the number of sensors. Simulation results show that the S₀-D+Seq(2-D) algorithm achieves a significant reduction in CPU times compared to the S-D algorithm with similar association qualities.