Clique-Based Location Estimations for Wireless Sensors in GPS-Free Environments

Abstract

In this paper we present a distributed, self-organizing, localization method for wireless sensors and a scheme for reducing position estimate errors by employing a novel concept of cliques of nodes. We describe an applicable theoretical foundation of peer-to-peer distance-vector (DV) exchange algorithm, which can operate without any manual pre-configuration. This method allows for propagating the complete information of a reference system (i.e., axes orientation and origin position) using angle and range measurements. In order to take advantage of possible spatial-diversity of measurements and to produce enhanced position estimates, we propose here an enhancement method which is based on groups of interconnected nodes (cliques) to combine localization measurements using a minimum-square-error (MSE) criterion and spatial transformations. This enables propagation of the reference system among all the nodes in a clique more accurately than by regular peer-to-peer distance-vector exchanges. Computer simulation reveals that the new clique based error mitigating scheme reduces the localization error by 65% for a typical underground tunnel setting.