Visible light communication based indoor localization

• Autor(a) principal: Chaudhary, Neha
• Data de Publicação: 2022
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/34991
• topic: Visible light communication; Visible light positioning; Indoor localisation; Artificial neural networks; Machine learning; Multipath reflections

The demand for highly precise indoor positioning systems (IPSs) is growing rapidly due to its potential in the increasingly popular techniques of the Internet of Things, smart mobile devices, and artificial intelligence. IPS becomes a promising research domain that is getting wide attention due to its benefits in several working scenarios, such as, industries, indoor public locations, and autonomous navigation. Moreover, IPS has a prominent contribution in day-to-day activities in organizations such as health care centers, airports, shopping malls, manufacturing, underground locations, etc., for safe operating environments. In indoor environments, both radio frequency (RF) and optical wireless communication (OWC) based technologies could be adopted for localization. Although the RF-based global positioning system, such as, Global positioning system offers higher penetration rates with reduced accuracy (i.e., in the range of a few meters), it does not work well in indoor environments (and not at all in certain cases such as tunnels, mines, etc.) due to the very weak signal and no direct access to the satellites. On the other hand, the light-based system known as a visible light positioning (VLP) system, as part of the OWC systems, uses the pre-existing light-emitting diodes (LEDs)-based lighting infrastructure, could be used at low cost and high accuracy compared with the RF-based systems. VLP is an emerging technology promising high accuracy, high security, low deployment cost, shorter time response, and low relative complexity when compared with RFbased positioning. However, in indoor VLP systems, there are some concerns such as, multipath reflection, transmitter tilting, transmitter’s position, and orientation uncertainty, human shadowing/blocking, and noise causing the increase in the positioning error, thereby reducing the positioning accuracy of the system. Therefore, it is imperative to capture the characteristics of different VLP channel and properly model them for the dual purpose of illumination and localization. In this thesis, firstly, the impact of transmitter tilting angles and multipath reflections are studied and for the first time, it is demonstrated that tilting the transmitter can be beneficial in VLP systems considering both line of sight (LOS) and non-line of sight transmission paths. With the transmitters oriented towards the center of the receiving plane, the received power level is maximized due to the LOS components. It is also shown that the proposed scheme offers a significant accuracy improvement of up to ~66% compared with a typical non-tilted transmitter VLP. The effect of tilting the transmitter on the lighting uniformity is also investigated and results proved that the uniformity achieved complies with the European Standard EN 12464-1. After that, the impact of transmitter position and orientation uncertainty on the accuracy of the VLP system based on the received signal strength (RSS) is investigated. Simulation results show that the transmitter uncertainties have a severe impact on the positioning error, which can be leveraged through the usage of more transmitters. Concerning a smaller transmitter’s position epochs, and the size of the training set. It is shown that, the ANN with Bayesian regularization outperforms the traditional RSS technique using the non-linear least square estimation for all values of signal to noise ratio. Furthermore, a novel indoor VLP system is proposed based on support vector machines and polynomial regression considering two different multipath environments of an empty room and a furnished room. The results show that, in an empty room, the positioning accuracy improvement for the positioning error of 2.5 cm are 36.1, 58.3, and 72.2 % for three different scenarios according to the regions’ distribution in the room. For the furnished room, a positioning relative accuracy improvement of 214, 170, and 100 % is observed for positioning error of 0.1, 0.2, and 0.3 m, respectively. Ultimately, an indoor VLP system based on convolutional neural networks (CNN) is proposed and demonstrated experimentally in which LEDs are used as transmitters and a rolling shutter camera is used as receiver. A detection algorithm named single shot detector (SSD) is used which relies on CNN (i.e., MobileNet or ResNet) for classification as well as position estimation of each LED in the image. The system is validated using a real-world size test setup containing eight LED luminaries. The obtained results show that the maximum average root mean square positioning error achieved is 4.67 and 5.27 cm with SSD MobileNet and SSD ResNet models, respectively. The validation results show that the system can process 67 images per second, allowing real-time positioning.