Перегляд за Автор "Kysil, Tetiana"

Зараз показуємо 1 - 2 з 2
  • Ескіз
    Документ
    Approach to a decentralized, physician-oriented ehr architecture with cryptographic protection
    (Хмельницький національний університет, 2025) Kysil, Volodymyr; Kysil, Tetiana
    In modern Electronic Health Record (EHR) systems, electronic records play a primary role in storing examination data; however, classic centralized storage systems face challenges regarding security, privacy, and data availability. Centralized databases are vulnerable to cyberattacks, data leaks, and interoperability issues with other systems, threatening patient confidentiality as well as the efficiency and transparency of medical institutions. In the specific case of a distributed network architecture where servers may operate autonomously without constant internet connectivity, decentralized solutions are required that combine cryptographic protection with ease of use for medical personnel. Physician-centric systems allow for workflow optimization in institutions where doctors collaborate within a trusted environment, while adhering to ethical standards of transparency for patients. The objective is to create a flexible, autonomous platform that ensures cryptographic data protection through envelope encryption with combined Data Encryption Keys (DEK), server key rotation, and hash chains for modification detection. The system supports profile migration between nodes, the exchange of signed data between physicians, and resource optimization by offloading completed records to a registry node. The primary focus is on patient transparency: any data decryption must be accompanied by a notification to the patient via an active notification system; if the notification system is unavailable, the decryption operation is not performed. Consequently, data decryption is logged (identifying the entity performing the decryption) for future reference. The system workflow begins with the creation of a patient profile on the physician’s local node, where data is encrypted using a combined DEK (static user key + daily node key). Hash chains ensure file integrity, preventing undetected changes, while two differently encrypted copies of the key allow for access recovery by an administrator. This makes the system resilient to failures, with minimal requirements for users, maintaining a balance between efficiency for medical personnel and patient rights regarding access information, data access, and data portability. Performance evaluation of the proposed architecture demonstrates that the storage overhead for cryptographic metadata is approximately 248 bytes per user profile (server-encrypted key, user-encrypted key, key-encrypted identifier, creation time; the size of encrypted data depends on the algorithm) and 40 bytes (SHA-256 standard + timestamp) per record to ensure cryptographic linkage. This is a negligible amount (<0.001%) compared to the volume of medical data (up to 32 MB per record with images). The absence of a need for global consensus (unlike blockchain solutions) ensures simple O(1) write operations, guaranteeing high performance even on resource-constrained hardware. Efficiency assessment indicates that the architecture adds minimal overhead: only a slight increase in size and access time within the constraints. Thanks to the local consensus model, the system does not require network synchronization, ensuring O(1) write complexity and high speed even on personal devices without dedicated server hardware.
  • Ескіз
    Документ
    Reinforcement learning method for autonomous flight path planning of multiple UAVs
    (Хмельницький національний університет, 2025) Velychko, Maksym; Kysil, Tetiana
    This study aims to develop a reinforcement learning method for autonomous flight path planning of multiple UAVs under real-world conditions with limited observations and multiple conflicting optimization objectives. The research proposes a multi-agent reinforcement learning approach based on Proximal Policy Optimization (PPO) combined with centralized training and decentralized execution (CTDE). Additionally, a recurrent neural network (RNN) layer is integrated into the critic and actor networks to address partial observability. The reward function is designed to balance time efficiency, safety, and area coverage. Experimental results demonstrate that the proposed method significantly outperforms independent learning approaches in terms of reward accumulation, convergence speed, and decision stability. The CTDE architecture with RNN-enhanced critics proved effective in handling the challenges of multi-agent coordination and partial observability. The trained model enables real-time trajectory planning in three-dimensional environments, surpassing traditional optimization methods. The novelty lies in the application of a multi-agent PPO architecture enhanced by RNNs under CTDE for solving real-time multi-objective optimization problems in UAV path planning. A customized reward structure was developed to simultaneously optimize safety, time, and coverage objectives without retraining. The developed method enables efficient and reliable online trajectory planning for UAV groups, making it applicable in surveillance, search and rescue, and exploration missions where rapid and adaptive decision-making is essential.