Swarm-Based Aerial Intelligence for Real-Time Data Integration in Digital Transformation Systems

Abstract

This paper presents a hybrid edge–cloud architecture for real-time aerial intelligence using autonomous drone swarms. The system combines onboard sensing and lightweight inference with centralized reasoning via large language models (LLMs) at the ground station and optional cloud-based analytics for strategic forecasting. Each UAV has multimodal sensors and performs edge-level preprocessing to reduce latency and bandwidth usage. Processed data is transmitted to a GPU-accelerated ground station, which executes LLM-based prompt interpretation, multimodal data fusion, sector classification, and geospatial threat mapping. The architecture supports real-time decision-making in mission-critical scenarios such as disaster response, environmental monitoring, and urban inspection.

Experimental simulations confirm that edge inference reduces response latency by up to 40%, with stable operation across mesh-based UAV networks up to 30 units. Bandwidth and latency were modeled under varying topologies and confirmed through throughput degradation curves and task saturation thresholds. Results validate that combining edge processing with cloud-enabled intelligence enables scalable, adaptive UAV operations aligned with the demands of digital transformation.