Received Signal Strength Indicator-Based Indoor Localization Using Nonlinear Dual Set-Membership Filtering

In most existing indoor localization techniques based on the received signal strength indicator (RSSI), the location accuracy and the impact of computational complexity on system performance are major concerns. To improve computational efficiency and reduce potential inaccuracies, the nonlinear dual set-membership filtering (NDSMF) is proposed. First, the critical parameters of transmit power and path loss exponent (PLE) are estimated by a multiobjective optimization algorithm in RSSI-based localization. Second, to avoid the errors and high computational complexity caused by the direct linearization of the nonlinear system and the multiple solutions of the semidefinite program (SDP) during the filtering iteration process, an NDSMF is designed based on the principles of strong duality and set theory to determine the ellipsoid set containing the optimal estimate of the target node. Then, based on the designed set-membership filter, a new ellipsoid-based fusion scheme is developed to prove that there exists a smaller and better-performing intersection ellipsoid set than all local ellipsoid sets. Finally, simulations and experiments are presented to validate the accuracy and effectiveness of the proposed algorithms. Under the same experimental scenario, the proposed algorithm achieves 47.6% and 58.9% improvement in localization accuracy compared with the currently mainstream nonlinear set-membership filtering (NSMF) and extended set-membership filtering (ESMF) algorithms, respectively.