Description
Date depot: 1 décembre 2023
Titre: Caractérisation des performances des LED pour la communication et la détection intégrées de la lumière visible
Directeur de thèse:
Xun ZHANG (LISITE)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Signal et communications
Resumé: The sixth generation of mobile communication (6G) represents the next crucial stage in the development of information and communication technologies, addressing the growing demand for higher-performance communication systems in future societies. 6G aims to harness the advantages of massive data and diverse application scenarios, constructing a seamlessly integrated wireless network. Through the integration of communication and sensing technologies, 6G envisions the creation of a more intelligent and adaptive communication system. In this context, Visible Light Communication (VLC) technology emerges as a key enabler for 6G communication systems, not only bringing new spectrum resources but also offering high-speed transmission and energy reuse features. VLC holds the potential to enhance the efficiency of communication networks, alleviate the increasing scarcity of spectrum resources, and promote the development of information and communication technologies towards a more carbon-neutral direction.LEDs, primarily designed as dedicated lighting devices, pose significant challenges for high-performance VLC applications due to their constrained modulation capabilities. To meet the demands of high-speed VLC, it is essential to model the dynamic nonlinear characteristics of LEDs and implement corresponding optimization measures to enhance their modulation bandwidth. The commonly employed Volterra model, known for its versatility in characterizing various nonlinear systems, has been widely used for LED modeling. In recent years, the surge in research on artificial intelligence and machine learning has led to their application in modeling and compensating for the nonlinear distortions in LEDs. However, existing LED models often treat LEDs as black boxes using complex fitting methods. Given the time-varying nature of LED nonlinearities and differences among different LEDs, establishing an LED equivalent circuit model based on their physical properties will aid in the parameter estimation process and reduce overall complexity. Furthermore, a semiconductor physics-based LED model can provide an effective reference for optimizing LED structures.While LEDs have a well-established evaluation system for illumination performance, including luminous efficacy, color rendering index, light decay, and color uniformity, there remains a significant gap in characterizing and evaluating the communication performance of various LED types. Despite numerous methods for optimizing LED bandwidth, traditional wireless communication devices commonly rely on the 3dB bandwidth as a metric for assessing communication capabilities. However, due to the bandwidth extension required for high-speed VLC systems based on LEDs, the 3dB bandwidth inadequately represents the potential communication capabilities of LEDs. This poses a considerable challenge to the development of LED-based VLC technology and its integration with existing wireless communication technologies. Therefore, analyzing the common characteristics of the frequency response based on a universal LED equivalent circuit model and proposing a characterization metric for the potential communication capabilities of LEDs will effectively guide LED optimization.For LED-based Visible Light Positioning (VLP), the key to determining user positions lies in the differentiation of information contributed by massive LED nodes. Current methods for node identification, such as beacon construction and signal fingerprint extraction, often require additional signals, leading to reduced spectrum efficiency and hindering true communication-sensing integration. Additionally, in RF positioning, researchers have successfully extracted transmitter fingerprints through received signal analysis without occupying any communication resources, albeit requiring a receiver with higher computational power. Therefore, we propose a fingerprint construction algorithm based on the LED equivalent circuit model. By extracting physical parameters from the equivalent circuit model, we aim to construct distinctive hardware fingerprint features, effectively promoting the integration of visible light communication and positioning.
Doctorant.e: Chen Xuanbang