MENA : Machine Learning Enhanced Next-Gen Smartphone Communication Augmented with Barcodes

Abstract

The rapid evolution of optical wireless communications (OWC), particularly visible light communication (VLC), has garnered significant attention due to advancements in light-emitting diode (LED) technology. Among the various subfields of VLC, optical camera communication (OCC) has emerged as a promising area of research, driven by the ubiquity of digital cameras in modern devices. This thesis presents the development of a novel short-range communication system, MENA, which leverages smartphone screens as transmitters (Tx) and cameras as receivers (Rx) to establish a screen-to-camera (S2C) communication link. The system encodes data into visual images displayed on the Tx, which are captured and decoded by the Rx over successive frames. By utilizing existing smartphone hardware without requiring any modifications, the proposed system offers a highly accessible and mobile solution for device-to-device communication.

The core innovation of this research lies in the integration of computer vision and machine learning (ML) techniques with VLC to enhance the robustness and efficiency of the S2C system. Specifically, convolutional neural networks (CNNs) are employed to improve the data transmission and reception.

The research thoroughly investigates the impact of various computer vision challenges on the system's performance, including fluctuating distances, orientations, and lighting conditions. Novel algorithms for blur reduction and correction are developed, alongside innovative CNN-based techniques to enhance synchronization, increase data throughput, and improve overall system performance. The study also conducts an in-depth analysis of transmitter behaviours, exploring the effects of different mobile devices, transmission codes, and augmented barcodes across various data packet sizes. The culmination of this research is the full implementation of the proposed S2C system on a mobile platform, followed by a series of comprehensive experiments to evaluate its performance. These experiments validate the system's potential, offering a robust analysis of its functionality across a spectrum of operational scenarios. The results affirm the viability and effectiveness of this novel S2C communication approach, paving the way for future applications in augmented reality, smart home technologies, and beyond.

Date of Award	30 Apr 2025
Original language	English
Awarding Institution	Northumbria University
Supervisor	Hoa Le Minh (Supervisor), Fary Ghassemlooy (Supervisor) & Wai Lok Woo (Supervisor)