Factor graph based multi-source data fusion for wireless localization

Multi-source fusion localization is an effective approach when a single source is unavailable or the positioning accuracy is unsatisfied, and it can take advantages of different location sources to achieve a better result. Data fusion is a process of merging different solutions and techniques with disparate types of information. In order to provide users with better location based services, this paper proposes a factor graph based multi-source data fusion algorithm for wireless localization. Different fusion sources are divided into multi-levels by adopting confidence level estimate algorithm. By using sum-product algorithm, the soft-information is calculated to complete the fusion process. Through some simulations, we can see that the proposed algorithm can improve the positioning accuracy greatly. At the same time, it has low complexity and a plug and play capability.