| Summary: | In most range-based localization methods, inferring distance from radio signal
strength using mathematical modeling becomes increasingly unreliable and complicated
in indoor and extreme environments, due to effects such as multipath propagation
and signal interference. We propose FuzLoc, a range-based, anchor-based,
fuzzy logic enabled system system for localization. Quantities like RSS and distance
are transformed into linguistic variables such as Low, Medium, High etc. by binning.
The location of the node is then solved for using a nonlinear system in the fuzzy
domain itself, which outputs the location of the node as a pair of fuzzy numbers. An
included destination prediction system activates when only one anchor is heard; it
localizes the node to an area. It accomplishes this using the theoretical construct of
virtual anchors, which are calculated when a single anchor is in the node’s vicinity.
The fuzzy logic system is trained during deployment itself so that it learns to
associate an RSS with a distance, and a set of distances to a probability vector.
We implement the method in a simulator and compare it against other methods like
MCL, Centroid and Amorphous. Extensive evaluation is done based on a variety of
metrics like anchor density, node density etc.
|
|---|
