Reconfigurable Intelligent Surfaces enabling Positioning, Navigation, and Timing Services
Md Sahabul Hossain, Nafis Irtija, Eirini Eleni Tsiropoulou, Jim Plusquellic, Symeon Papavassiliou
Abstract
In this paper we exploit the advances provided by the Reconfigurable Intelligent Surfaces (RIS) technology, which allows for the software-defined control of the electromagnetic properties of the wireless medium, in order to introduce a low-cost and easily deployable ground-based alternative positioning, navigation, and timing (PNT) service. According to the proposed solution, the positioning of the target is performed by one original signal transmitted by one base station (BS), and three additional signals reflected by three RISs (selected from a set of available RISs), thus reducing significantly the required implementation and infrastructure cost. Different reinforcement learning algorithms, based on the gradient ascent and the log-linear learning models, are introduced and investigated to enable the target, to autonomously and dynamically select the optimal set of three RISs to be used for the optimization of its positioning accuracy. Subsequently, an iterative least square algorithm is realized to determine the position of the target. Detailed numerical results are presented that highlight the tradeoffs of the introduced reinforcement learning algorithms in terms of convergence and impact on achieved positioning precision, and demonstrate the superiority of the proposed methodology against existing ground-based PNT solutions.