Hybrid initialization for non-convex network localization problems

摘要: Solving the distance-based network localization problem typically entails formulation of an equivalent optimization which is either convex but sub-optimal, or optimal (i.e., maximum-likelihood) non-convex. We show that non-convexity implied by choice need not be translated onto high computational complexity nor to performance degradation. To this end, we focus on approach whereby low-complexity algorithms are coupled with efficient initialization in turn, formulated terms Euclidean Distance Matrix (EDM) completion under condition percolated (as required Graph-based Completion). The resulting Hybrid Initialization scheme shown sufficient bring such as SMACOF and NLS close far more sophisticated alternatives SDP.