New graph-based and MCMC approaches to multi-INT surveillance

We propose a graph-based (i.e. pairwise-cost) formulation to the multi-INT track fusion problem. In addition to the kinematic Markov (path-independence) assumption, we introduce the assumption that the impact of emitter tracks is solely to identify feasible kinematic tracks for association. Subject to these assumptions, we are able to formulate an integer linear programming (ILP) problem that is of much smaller dimension than the one in an MHT-based solution. Performance results suggest that the approach is quite promising. Further improvements to the multi-INT graph-based tracking solution may be achieved with MCMC-based statistical sampling.