Authors: Vahideh Moghtadaiee; Nasim Alikhani; Seyed Ali Ghorashi
Addresses: Cyberspace Research Center, Shahid Beheshti University G. C., Tehran, 1983963113, Iran ' Cyberspace Research Center, Shahid Beheshti University G. C., Tehran, 1983963113, Iran ' Cognitive Telecommunication Research Group, Department of Electrical Engineering, Shahid Beheshti University G. C., Tehran, 1983963113, Iran
Abstract: One popular indoor localisation method in wireless sensor networks (WSN) and wireless local area networks (WLAN) is fingerprinting technique. However, the training stage of fingerprinting and constructing the radio map is time-consuming and labour-intensive. Here, we propose a novel method to decrease the training cost by building a simulated radio map using an improved path-loss model in which the impacts of line of sight (LOS) and non-light-of-sight (NLOS) propagations are considered. Including LOS/NLOS effects also improves distance estimation. Furthermore, the simulated radio map helps assess and improve the fingerprinting area and network parameters prior to the actual positioning. For performance evaluation, three different radio maps are created using simulation and a test field experiment and then compared based on deterministic and probabilistic algorithms. The results indicate that the improved model outperforms the typical path-loss model and the localisation error gets closer to the actual error of the fingerprinting network.
Keywords: indoor localisation; fingerprinting; RSS; received signal strength; LOS; line of sight; NLOS; non-light-of-sight.
International Journal of Sensor Networks, 2020 Vol.32 No.1, pp.15 - 24
Received: 31 Jul 2019
Accepted: 01 Aug 2019
Published online: 12 Jan 2020 *