Sensing-Aided 6G Drone Communications: Real-World Datasets and Demonstration

Abstract

In the advent of next-generation wireless communication, millimeter-wave (mmWave) and terahertz (THz) technologies are pivotal for their high data rate capabilities. However, their reliance on large antenna arrays and narrow directive beams for ensuring adequate receive signal power introduces significant beam training overheads. This becomes particularly challenging in supporting drone communication, where the dynamic nature demands frequent beam alignment to maintain connectivity. Addressing this bottleneck, our paper introduces a machine learning-based framework that leverages multi-modal sensory data, including visual and positional information, to expedite mmWave/THz beam prediction. Unlike conventional approaches that depend on exhaustive beam training, our solution incorporates contextual data to accurately predict beam directions, significantly mitigating the training overhead. Additionally, our framework predicts future beam alignments ahead of time, enhancing the system’s responsiveness by addressing challenges posed by drone mobility and computational delays. We validate our approach through evaluations on a real-world mmWave drone communication dataset, integrating camera visuals, GPS coordinates, and mmWave beam training data. Our findings demonstrate a top-1 beam prediction accuracy of 86.32%, with near-perfect top-3 and top-5 accuracies. For future beam predictions, our vision-aided solution achieves notable accuracies of approximately 92% and 88% for predicting the second and third future beams respectively, using the top-3 accuracy metric. These results not only underscore the efficacy of our sensing-aided solution but also mark a significant stride towards realizing efficient and highly-mobile 6G drone communications, potentially transforming future aerial networks.