Location sensing with robust minimax thin-plate spline

In this paper, we propose RMTS, a localization algorithm based on the concept of robust minimax thinplate spline for indoor location sensing. Short-term average of received signal strength (RSS) measurements from a transmitter node are modeled as a function f : Rd × Rd ! R subject to certain convexity constraints. Localizing a target essentially translates into finding the inverse given measured RSS values at unknown locations. Unlike existing mapping approaches, RMTS does not require extensive site survey in RSS map construction. Empirical results demonstrate that RMTS outperforms existing methods and has comparable or better performance than a genie-aided Bayesian Inference method.