An Enhanced ZigBee Indoor Positioning System With an Ensemble Approach

This paper presents a framework for ZigBee indoor positioning with an ensemble approach. This approach exploits the complementary advantages of various algorithms, weights the estimation results, and combines them to improve accuracy. This is achieved by dynamically analyzing the diverse patterns of inputs and combining base positioning algorithms with spatial dependent weights. The experiments were conducted in a realistic ZigBee sensor network. Results demonstrated that the proposed approach apparently achieves more accurate location estimation than the compared methods including the gradient-based search, linear squares approximation, multidimensional scaling, fingerprinting method, and a multi-expert system.