Vehicular and Positioning Visible Light Systems with Low-cost LED-based Devices

Artificial lighting is everywhere, from the light bulbs on our ceilings to car headlights. This makes Visible Light Communication (VLC) an attracting technology since it uses modulated optical radiation in the visible light spec- trum, exploiting the light emission of LEDs. The main advantage of VLC is that it can provide both standard illumination and data connection. Also VLC systems are suitable for many different applications due to their rela- tively simple design for basic functioning, efficiency and large geographical distribution. Recently both Vehicular VLC (V 2LC) and LED-based Posi- tioning Systems (LPS) are gaining a lot of attention from industry and the scientific community due to their great potential in terms of driver safety and positioning accuracy, respectively. The aim of this PhD dissertation is to investigate two different VLC systems: a V2LC system and an LPS technique. Both are designed for using low-cost components and for being analyzed in realistic environments by exploiting two dedicated testbeds. An LED-based traffic light for sending traffic and safety information, and a 2D localization algorithm implemented on an open source platform (OpenVLC), are studied. For the V2LC topic, after a preliminary study of a simple propaga- tion VLC channel, an IEEE 802.15.7-complaint low-cost software-defined transceiver along with an experimental VLC application are presented. The transceiver implements the specific standard PHY-level for outdoor VLC and it is able to stream data at 100 kbit/s. For the positioning system, a new multipath detection technique for filter- ing out reflections and better positioning with light is designed and tested in realistic scenarios. More in details, the multipath detection does not require the knowledge of the channel impulse response, and that it is suited to be implemented in low-cost positioning receivers that use a single photodiode. To develop this technique, (i) the statistical properties of Non-Line-of-Sight (NLOS) components is analyzed, (ii) an automated testbed to study the re- flections of different types of surfaces and materials is developed, and (iii) an algorithm to remove the NLOS components affecting the positioning is designed. Experimental evaluation shows that in complex environments this methodology can reduce the localization error using a single photodiode up to 93%.