Suboptimal Position Control as Enabler for Low-Cost Distance Estimation in Dynamic Multipath Networks

Konferenz: WSA & SCC 2023 - 26th International ITG Workshop on Smart Antennas and 13th Conference on Systems, Communications, and Coding
27.02.2023–03.03.2023 in Braunschweig, Germany

Tagungsband: ITG-Fb. 308: WSA & SCC 2023

Seiten: 5Sprache: EnglischTyp: PDF

Autoren:
Kokorsch, Marcel; Dietl, Guido (Julius-Maximilians-Universität Würzburg, Insitute of Computer Science, Professorship of Satellite Communication and Radar Systems, Würzburg, Germany)

Inhalt:
Low-cost drones, such as those oftenly used in drone swarms, present certain challenges due to the limited hardware available. Due to this reason and the fact that it cannot be used indoor, the Global Positioning System (GPS) is not feasible in order to localize the individual drones in a swarm. In such scenarios, multilateration using the drones as well as some anchor nodes whose positions are known is a common localization method. However, this approach depends on knowing the distances between the drones. This is the reason why we focus on radio based distance estimation between two drones using the Received Signal Strength Indicator (RSSI) as well as a basic path loss model here. However, in multipath environments, this model is no longer valid. Moreover, radio based distance estimation leads to ambiguities which degrades tremendously the accuracy of the localization method based thereon. In this paper, we propose an estimation method which is introducing position variations in order to mitigate interference caused by multipath propagations. To do so, averaged RSSI measurements are considered. In case of low-cost drones, these position variations do already exist since most of these drones have an inaccurate position control. Finally, the basic path loss model is used for distance estimation. Particularly promising are the facts that this approach works without requiring communications overhead or significant computational power on the drones. Monte Carlo simulations show that the empirical mean square error of the distance estimates can be drastically reduced using the proposed approach. These results are further confirmed by measurements in the laboratory.