Beamforming innovations for wireless communication and optical computation

Abstract

Beamforming, originally developed for radar applications during the mid-20th century, has evolved significantly to become a cornerstone in modern communication and signal processing technologies. This advanced technique enhances signal directionality and focus, which is crucial for minimizing interference and maximizing the efficiency of transmission systems. This dissertation investigates the extensive applications and transformative potential of beamforming in contemporary technological fields. This investigation particularly focuses on its integration into wireless communication systems and optical computing. The first major application discussed is Transfer Beamforming (TBF) for wireless communication systems, where it revolutionizes the management of signal energy, enhancing data transmission range and system reliability in densely populated tag environments like warehouses. By leveraging semi-active tags as initial sniffing tools, TBF facilitates the beamforming process to transfer energy effectively to passive tags, substantially increasing their reading range and reducing miss reading rates. Prototype testing in a crowded warehouse achieved a 99.9% inventory coverage rate, demonstrating a power transmission improvement of 6.9 dB and a doubling of inventory speed compared to existing methods. Additionally, the dissertation extends the application of beamforming principles to the realm of optical computing, specifically through the development of Binary Optical Neural Networks (BONNs). By leveraging the beamforming characteristic, BONNs selectively illuminate specific regions when processing different data inputs, thereby effectively accomplishing classification tasks. BONNs employ binarized weights for low-cost fabrication and are capable of processing large-scale data with significantly lower energy requirements. Prototypes show that BONNs consume up to 2,405 times less energy than conventional Electronic Neural Networks(ENNs) while maintaining an average recognition accuracy of 74% across various datasets. The reduction in layer manufacturing costs to just 0.13 per layer presents a scalable, cost-effective alternative to traditional computational models. This dissertation demonstrates the impact of beamforming innovations on both wireless communication systems and optical computation, reflecting a significant evolution of beamforming. Through the implementation of Transfer Beamforming (TBF) in dense wireless environments and the development of Binary Optical Neural Networks (BONNs), this research highlights the adaptability and efficiency of beamforming techniques in modern technological applications. TBF's ability to enhance data transmission accuracy and the potential of BONNs in processing large datasets with minimal energy displays beamforming's potential to revolutionize communication and computational paradigms. These advancements not only improve system efficiencies but also pave the way for future applications, solidifying beamforming's role as a cornerstone technology that bridges the gap between traditional signal processing and next-generation network and computational technologies.