Constrained decentralized algorithm for the relative localization of wearable wireless sensor nodes

Motion capture might become one key feature of future Wireless Body Sensor Networks (WBSN), allowing new applications such as home activity monitoring, nomadic postural rehabilitation or sportive gesture recording through standard on-body communications. In this context we present herein a stand-alone solution that enables the localization of wearable wireless nodes relatively to a body-strapped Local Coordinate System (LCS). In particular, we consider adapting a Distributed Weighted Multi-Dimensional Scaling (DWMDS) algorithm fed by cooperative inter-node range measurements obtained through e.g., Time Of Arrival (TOA) estimation, where estimated nodes' locations are asynchronously updated based on their neighborhood information. Exploiting further the presence of constant-length radio links, a Constrained solution (CDWMDS) is thus proposed to improve localization accuracy, while reducing traffic and power consumption. Another novelty lies in the initialization step, which somehow benefits from the space-time correlation of nodes' locations under body mobility. Relying on a realistic biomechanical model, we provide preliminary simulation results to illustrate the relative gains observed in comparison with a nominal algorithm setting.