Indoor Wi-Fi/PDR Fusion Localization Based on Adaptive and Robust Kalman Filter

In response to the problem of dynamic channel state information in complex indoor environment, this paper proposes an adaptive and robust Kalman filter approach for indoor Wi-Fi/Pedestrian Dead Reckoning (PDR) fusion localization. This approach conducts the multiple location information fusion of Wi-Fi propagation model and PDR to infer the optimal location estimate of the user. At the same time, based on the filter feedback mechanism, the fusion localization result is used to dynamically modify the path loss exponent in weighted least square method as well as the observation covariance in filter model with the purpose of guaranteeing that the Wi-Fi propagation model is close to the real indoor environment. The experimental results indicate that the proposed method is capable of well solving the problems of low localization accuracy by using the Wi-Fi solely and accumulative error in PDR. Furthermore, the real-time modification of path loss exponent and observation covariance improves the stability of the proposed fusion localization system. © 2019, Chinese Institute of Electronics. All right reserved.