Abstract
Given their considerable environmental and economic impact, the precise and prompt identification and prediction of forest fires are paramount. This study introduces an innovative system enabling swift and protected data transfer from low-orbiting satellites, crucial for live forest fire observation. Our novel framework integrates the BB84 quantum key distribution protocol to ensure communication security and employs convolutional neural networks (CNNs) for advanced processing of Earth remote sensing (ERS) data. These components are cohesively managed within a singular geographic information system (GIS). Experimental data show that our approach maintains a 100 Mbps data transfer rate with minimal errors, even amidst noisy channel interference. Additionally, the BB84 protocol encrypts a gigabyte of data in approximately 5.3 seconds. Notably, CNN architectures like EfficientNet-B0 demonstrated high detection accuracy, reaching 94.1% even with 10% noise in the data. Collectively, these outcomes underscore the system's robust performance in increasing the precision of forest fire detection, safeguarding data integrity, and enabling immediate operational use.