stochastic network interdiction model for human trafficking intervention

• Online Access: http://hdl.handle.net/2047/D20317903

Network interdiction models are widely used in security and resilience analysis applications. In this thesis, we create a network interdiction model for human trafficking interventions. Our model considers law enforcement's detection and intervention of human trafficking as interdictions, interdictions are binary and their effects are stochastic (i.e. there is a positive probability that interdiction is unsuccessful). The problem which the follower (trafficker) tries to solve is a maximum flow problem with the capacity parameter serving as a proxy for path desirability. Therefore the objective for the leader (interdictor) is to minimize the expected maximum flow of the trafficker. The multi-stage version of our model considers gaining information from successful interdictions which affects interdiction probabilities in later stages. We created a county level and a town level example network for the state of Massachusetts by using census data and solved the model on these networks by using a genetic algorithm. Finally, we discuss our observations about the performance of the algorithm on the above mentioned networks.