To address the problems of poor generality of indoor pedestrian localization and tracking schemes and low accuracy in complex environments, He Huang's research team from China proposed a new indoor localization scheme based on IMU (inertial measurement unit) and Wi-Fi RSS - VariFi, which combines variational Bayesian inference and gradient descent algorithm, which optimizes the inference network parameters in real time based on reconstruction loss and regularization loss, and thus obtains an accurate estimated position.

The team conducted static positioning and trajectory estimation experiments to evaluate the performance of VariFi. The experimental results show that VariFi's positioning accuracy is 39.82% higher than mainstream positioning solutions, which not only enables high-precision positioning, but also shows good stability.