PROTECTED LOCAL ACCESS BASED ON PHYSICALLY-DETERMINISTIC LINKS

Abstract

This article proposes an approach to organizing protected local access based on physically-deterministic links for intelligent medical diagnostic systems. The relevance of this study is driven by the necessity of ensuring confidential remote access to local computing resources operating within isolated networks or behind restrictive systems, such as NAT. The developed approach is based on the concept of a "blind" registry and out-of-band key transmission via physical media, such as QR codes or their analogs, enabling a zero-knowledge access model. Information interaction is protected by implementing physically-deterministic links that combine temporary tunnels with a mechanism for local metadata decryption in the user's browser using the Web Crypto API.This approach eliminates the possibility of data compromise at the network infrastructure provider level and ensures confidentiality without requiring centralized key management systems (KMS), which is vital for autonomous nodes and ease of use. The practical significance of the method is integrated into the general concept of automated diagnostic and consultation information technology, providing secure remote calls to neural network models deployed on local computers or servers. The proposed solution is the subject of ongoing research aimed at minimizing hardware requirements while maintaining high levels of access simplicity and setup for patient data access