A RVM-based RSS model for device-free localization

Radio tomographic imaging (RTI) is an emerging technique of device-free localization (DFL). The main challenge of RTI is the multipath interferences in RSS measurements, which could make the links become more unpredictable and finally lead to unsatisfactory DFL performance. For addressing this challenge, this paper presents a novel modeling method based on relevance vector machine (RVM), which can obtain the weight parameters of RSS model by using the training data from extensive experiments. There are two contributions in this paper. On one hand, the proposed modeling method can realize a two-dimensional parametrization of RSS variations in terms of both the excess path length and the proximity to TX/RX nodes. On the other hand, the RVM-based method is complementary to theoretical or empirical modeling methods and has the advantages of powerful data representation capability and better adaptability. The two-dimensional model is experimentally validated in the context of RTI and the results fully prove the superiority of the model.