We present a localization algorithm for low-power long-range Internet-of-things (IoT) networks, which exploits angle of arrival (AoA) and range information from non-coherent received signal strength (RSS) measurements. In this work, each anchor node is equipped with array antennas of known geometry and radiation patterns. The position of the target node and the path-loss exponent to each anchor are unknown and possibly time-varying. The joint estimation problem is formulated with a Bayesian model, where the likelihood functions are derived from the classical path-loss model and an RSS difference model. A message passing method is then exploited for efficient computation of the marginal posterior distribution of each unknown variable. The proposed algorithm is validated using real outdoor measurements from a low-power wide area network based IoT system in a challenging scenario. Results show that the proposed algorithm can adapt to dynamic propagation conditions, and improve the localization accuracy compared to a method that exploits only single geometric feature. Furthermore, the algorithm scales well in different antenna array configurations, and is compatible with various existing IoT standards.
|Titel||2020 54th Asilomar Conference on Signals, Systems, and Computers|
|Erscheinungsort||Pacifc Grove, CA, USA|
|Publikationsstatus||Veröffentlicht - 1 Okt 2020|