Design, modeling, and analysis of visual MIMO communication.

摘要

Today's pervasive devices are increasingly being integrated with light emitting diode (LED) arrays, that serve the dual purpose of illumination and signage, and photo- receptor arrays in the form of pixel elements in a camera. The ubiquitous use of light emitting arrays (LEA) and cameras in today's world calls for building novel systems and applications where such light emitting arrays can communicate information to cameras. This thesis presents the design, modeling and analysis of a novel concept called visual MIMO (multiple-input multiple-output) where cameras are used for communication. In visual MIMO, information transmitted from light emitting arrays are received through the optical wireless channel and decoded by a camera receiver. The paradigm shift in visual MIMO is the use of digital image analysis and computer vision techniques to aid in the demodulation of information, contrary to the direct processing of electrical signals as in traditional radio-frequency (RF) communication.;The unique aspect of camera communications is that visual perspective distortions dominate over distance-based attenuation, multipath fading and other important properties of the radio-frequency (RF) wireless channels. In visual MIMO, camera receivers together with LEAs allow multiple parallel channels as in RF MIMO to achieve throughput gains, but these gains depend on the perspective---orientation and distance---between the transmitter and receiver. Camera receivers also allow for a large field-of-view for signal reception and can facilitate tolerating mobility through intelligent tracking techniques to locate the light emitting transmitter in view. This dissertation studies these key aspects of visual MIMO communication, and has been structured into three parts. The first part derives the perspective dependent channel model and information capacity of visual MIMO communication, along with a case-study of capacity of camera communication using display screens as transmitters. Part two discusses perspective dependent throughput enhancement techniques that exploit the MIMO array structure and uses vehicle-vehicle (V2V) communication as a running example. Finally, part three discusses transmitter localization techniques that help adapt to mobility in visual MIMO channels. The inferences and lessons learned through this thesis open up novel opportunities to use cameras as an integral part of a communication system. Efforts have already been initiated by the optical wireless communication community to standardize camera communications, and such advances attest to the importance of using cameras for communications.