Visible Light Communications Indoor Sensing and Channel Modeling

Visible light communications (VLC) is considered a prime candidate for indoor access networks due to its many advantages, such as high physical layer security, massive access support, and high multiuser throughput. However, to achieve all the VLC systems' potential, a reconceptualized design procedure is required at all system levels. This dissertation describes the research and industry motivation for employing VLC as a promising future indoor network access, the fundamental research gaps that need to be addressed, and proposes some methods in different parts of VLC systems, including localization, sensing, and communication applications.

We introduce a VLC-specific indoor positioning system that estimates users' locations on the network-side, exploiting the uplink channel characteristics. In other approaches, multipath reflection degrades the performance of visible light communications (VLC) based localization systems, where it is often considered as a strong random noise. However, due to the inherent transmission features of light, the optical wireless indoor channel is static; therefore, multipath components can be modeled as deterministic functions of the transceiver location, furnishings, and room geometry. We exploit this unique feature of optical wireless communication channel for localization and investigate the performance limits of fingerprinting-based localization with multipath reflection as a source of information, i.e., a fingerprinting map. Limits on the localization accuracy are determined using the Cramer-Rao lower bound (CRLB) for different numbers of photodetectors deployed in the system and received signal features captured. The tightness of the analytical CRLB is tested by comparing it to the performance of a fingerprint-based positioning algorithm that uses the nearest neighbor method. Simulation results show an achievable root mean squared positioning accuracy of 45 cm and 5 cm (for one and four photodetectors, respectively), for an empty room. We then investigate the practical limitations on localization accuracy caused by a narrow transceiver bandwidth. Numerical results show that the localization system can still achieve decimeter accuracy for system bandwidth of 200 MHz, which makes fingerprinting schemes practical for off-the-shelf infrared devices.

The performance of VLC systems is vulnerable to the line of sight (LOS) link blockage due to objects inside the room. Considering pedestrians as the most common VLC links blocking obstacles, we develop a probabilistic passive pedestrian detection and localization method. Our method takes advantage of the blockage status of VLC LOS links between the user equipment (UE) and transceivers on the ceiling to passively detect a single pedestrian, modeled as a cylinder with a random radius. The VLC network gathers the blockage status and computes the geometry of the LOS link graph through a cooperative scheme between VLC device-equipped users inside the room. We also develop a mathematical framework to obtain an optimum solution for estimating the location and size of the object and conclude with a sub-optimum estimation by simplifying the problem to a quadratic programming approach. Simulation results show that using a 5 x 5 grid of transceivers on the ceiling and as few as eight UEs, the root-mean-squared error in estimating the center and radius of the object can be less than 5 cm and 3 cm, respectively.

We also develop a model for the indoor channel where the channel impulse response can be learned depending on available information, such as pedestrian localization, tracking, and room geometry. The blockage effect is one of the most critical concerns in optical wireless communication (OWC) systems, where one or more obstructions block the LOS signal path, causing the system performance to degrade drastically. To address this issue, we introduce a probabilistic hierarchical model of shadowing that relies on prior knowledge of object uncertainties such as its dimensions, as well as practical uncertainties like mobility and object detection error. Numerical results presented for both the LOS and non-line of sight (NLOS) parts of the received OWC signal in indoor environments consider a pedestrian as the main moving object. The result shows a maximum standard deviation of 10 nW from the randomness in the object's reflection, using the range of human body dimensions, when the location of the object is detected accurately; the variability can increase by orders of magnitude considering mobility uncertainty as well as object detection error. The model presented also enables the network to predict LOS blockage based on perfect knowledge of the object location. We then explain how this accurate channel model can contribute to taking advantage of the spatial resolution of light, therefore reaching the potential throughput of VLC.