Track Correlation Algorithm Based on Neural Network

In a distributed multi-sensor fusion system, the generalized classical assignment association algorithm is a minimum problem with constrains. A neural network scheme for track correlation problem is proposed to avoid exponential increase of computational complexity with increase of dimensions. In order to utilize the ability of Hopfield for combinatorial optimization problems, a multiple targets energy function is constructed to deal with constrained integer programming. Neural network is a sort of parallel approach. Hence its computational time will not increase exponentially with the increase of dimensions, and the complexity is obviously reduced. Finally, simulation results are given and show the validity of the proposed scheme.