Nonparametric belief propagation (NBP) is well-known probabilistic method for cooperative localization in sensor networks. However, due to the double counting problem, NBP convergence is not guaranteed in the networks with loops or even if NBP converges, it could provide us less accurate estimates. The well-known solution for this problem is nonparametric generalized belief propagation based on junction tree (NGBP-JT). However, there are two problems: how to efficiently form the junction tree in an arbitrary network, and how to decrease the number of particles while keeping the good performance. Therefore, in this paper, we propose the formation of pseudo-junction tree (PJT), which represents the approximated junction tree based on thin graph. In addition, in order to decrease the number of particles, we use a set of very strong constraints. The resulting localization method, NGBP based on PJT (NGBP-PJT), overperforms NBP in terms of accuracy and communication cost in any arbitrary network.