Opportunistic Wireless Localization System (OWLS) based on SLAM
Date
2013-12-04
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Widely accessible GPS fails in indoor environments and dense urban areas due to low SNR, multipath propagation, and LOS blockage. Taking advantage of signals of opportunity from locally generated wireless networks such as WiFi, WLAN, or 4G LTE is considered as a solution to ameliorate this issue. However, the mobile receiver must deal with reception of a multitude of disparate signals with unknown and random parameters. A systematic simultaneous localization and mapping (SLAM)-based approach is proposed to incorporate all information regarding the large number of source and channel unknown parameters along with the mobile node (MN) location. The Bayesian Fisher information matrix (BFIM) is derived as a central key to the SLAM-based opportunistic wireless localization system (OWLS) that unifies all information from observables, the MN trajectory, and a priori knowledge in one single matrix. The BFIM illustrates that observability of a SLAM-based OWLS is achieved under the assumptions of stationary ANs and the MN smooth trajectory.
Two Bayesian solutions based on the particle filter (PF) and the extended Kalman filter (EKF) are presented for opportunistic wireless localization to track MN locations and to incrementally build the map of the environment while simultaneously using this map to update the MN location(s). It is shown that the range offset due to line-of-sight blockage and MN clock drift leads to a large localization and mapping error. To enhance the tracking performance in non-stationary line-of-sight/non-line-of-sight (LOS/NLOS) propagation, the OWLS is modeled as a jump Markov nonlinear process where the source sight state is the jumping feature that derives the system dynamics in LOS or NLOS conditions. To jointly track the discrete state of sight condition along with other continuous location variables, local maximum likelihood and interacting multiple model estimators are fused with the previously developed PF-based and EKF-based solutions. A comprehensive performance analysis is carried out for the proposed OWLS and its solutions by exploiting theoretical bounds of error obtained from the BFIM, as a benchmark to assess the simulation results. It is shown that jump Markov Bayesian solutions for the SLAM-based OWLS mitigate both the clock drift and NLOS effects.
Description
Keywords
Statistics, Engineering--Electronics and Electrical, Robotics
Citation
Moazen Chaharsoughi, N. (2013). Opportunistic Wireless Localization System (OWLS) based on SLAM (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/27204