After a decade of developments in wireless sensor network (WSN) systems, the range of WSN applications is growing. The localization of static or mobile sensor nodes is an important application service required in WSNs. The problem becomes more challenging in indoor environments. This paper presents an indoor mobile sensor node localization method (using static sensor node infrastructure) using a fuzzy logic model to deal with uncertainties in range measurements (range-based localization). The authors claim that their localization method provides more robustness in position estimation; is fault tolerant and applicable in low-density WSNs; and performs better than some existing methods in the presence of range uncertainties.
The paper first presents results from detailed experiments measuring inter-node received signal strength indicators (RSSIs) and link quality indicators (LQIs) at different indoor pairwise node distances. The authors show that the derived statistical sensor model contains uncertainty in position estimates, and thus requires fuzzy logic estimation. The fuzzy logic estimation problem calculates a fuzzy density or fuzzy belief over the space of all locations. The fuzzy belief function is estimated using a standard predict-update cycle of recursive state estimators. The prediction step consists of the proposed dilation process, and the update step involves intersecting the predicted belief with beliefs induced by all inter-node position measurement observations. The authors then propose an improvement to the resulting position estimator by adding space constraint information using a Voronoi diagram or Dirichlet tessellation method. The experimental evaluation is performed in an indoor environment with Tmote Sky sensor nodes as the static sensor nodes and an on-board Pioneer 3-AT mobile robot. To compare the proposed method with a variant of the Monte Carlo localization method, the authors perform kidnapping and tracking experiments.
Overall, in the presence of uncertainties in range measurements, the idea of using a fuzzy logic estimator for localization is useful. This well-written paper is a good read for those interested in indoor localization research problems.