A multisensor multitarget data association algorithm

An algorithm for solving the following problem is presented. A set of three spatially distributed passive sensors and active 2-D radars is given. Passive sensors measure the azimuth and elevation angles of the source, while the active 2-D radars measure the range and the azimuth angle of the source. The source can be either a real target, in which case the measurement is the true measurement of the target plus measurement noise, or a false alarm. The sensors have nonunity detection probabilities. The measurements from the three sensors are associated to identify the real targets and to obtain their position estimates. Mathematically, the measurement-target association problem leads to a generalized three-dimensional (3-D) matching problem. The algorithm consists of two distinct phases: (1) costs are assigned to each possible measurement-target associations; and (2) a feasible set of associations is found such that the cost of association is minimized. The above NP-hard 3-D matching problem is solved via a series of polynomial-time weighted bipartite matching problems using an iterative Lagrangian relaxation technique. The matching algorithm produces a feasible solution at every iteration and also provides a tight bound on the optimality of the feasible solutions