Refinement of weighted centroid localization using a regular infrastructure topology

Abstract

Received signal strength (RSS) readings of common RF transceivers as a low-cost range sensor metric enable a reliable indoor local positioning system (ILPS). The mobile blind nodes (BNs) send out RF signals which are received and evaluated for the RSS by stationary reference nodes (RNs). The low-complexity weighted centroid localization (WCL) algorithm is a suitable estimator for the erroneous RSS-based range estimates. In many cases, the RNs are installed in a regular manner, e.g. at the walls and with regular node distances. The novel approach is to convert the information of this regular topology into an increased positioning accuracy. For a given RN topology, the systematic error of WCL can be computed beforehand and used to compensate the systematic error during the online positioning. The accuracy of WCL estimates is improved significantly, especially at the edge of the RN grid. The basic WCL and the WCL using the proposed topology refinement technique are compared for a 2D positioning in a factory hall environment. The refinement is most effective for infrastructure setups with very limited number of reference nodes, enabling a cost effective and reliable positioning solution.