Convolutional Neural Network-based Optical Camera Communication System for Internet of Vehicles (1911.09529v1)

Abstract: The evolution of internet of vehicles (IoV) and the growing use of mobile devices with the development of the Internet of Things, demand has grown for alternative wireless communication technologies. As a promising alternative, optical-camera communication (OCC) has emerged that uses light-emitting diode (LED) and camera as transmitter and receiver respectively. Since LEDs and cameras are exploring in traffic lights, vehicles, and public lightings, OCC has the potential to handle the transport systems intelligently. Though some technologies have been proposed or developed, these are not mature enough to uphold the huge requirements of IoV. However, most of the OCC applications are limited to single vehicle and there has been limited focus on the use of multiple vehicles detection (spatial) or fast processing (temporal) systems. Also, there has no system to challenge with the bad weather condition. In this research, a vision camera and high-speed camera has been proposed to provide multiple vehicle detection and fast data processing. Here, a convolutional neural network (CNN) based vehicular OCC system has been introduced to guarantee communication to maintain communication at the adverse condition and near-infrared is proposed in addition to visible lights to provide long range and secure communication. To support IoV communication, software-defined networking (SDN) has been proposed. Finally, the results represent the required conditions for vehicular OCC system analysis and improved performance demonstration using the proposed intelligent system.