Message Passing Based Wireless Multipath SLAM With Continuous Measurements Correction

The core of multipath-based simultaneous localization and mapping (SLAM) is to utilize the multipath propagation of signals to simultaneously achieve the estimation of the user and the surrounding environment's states. Existing multipath-based SLAM methods are Bayesian estimators that use the current-snapshot signal as input to update the states. However, the internal correlation of time-varying signals is neglected. To utilize the time sequence information in state estimation, we propose a new Bayesian model considering historical measurements and represent it by a factor graph. Based on the model, we develop BP-CC (Belief Propagation with Continuous measurements Correction), a message passing-based algorithm, considering the measurements of the previous snapshot to jointly estimate the agent's states and its surroundings. Numerical simulation results show that the proposed SLAM algorithm with continuous measurements correction, compared to the state-of-art algorithms, significantly increases the estimation accuracy and reduces the complexity with a better convergence ability.