Cyber-Physical System Architectures for Dynamic, Real-Time “Need-to-Know” Authorization

Abstract

Discussed is an Intelligent Information Agent (I ² A) architecture for real-time, adaptive, need-to-know, authorization of access to confidential/classified information. The “Need-to-know” paradigm for information content access and access to application and service execution is desperately needed for the ubiquitous non-traditional physical computing systems (e.g. refrigerators, wearable computers, engine sub-systems etc.), beginning to connect into the cyber mesh. The multi-agent system is based on the ELYSE cognitive neural, intelligent agent framework and provides “need-to-know” context-based authorization of requests for access to confidential/classified information. “Need-to-know” authorization is that which grants access to confidential/classified information only if that information is necessary for the requestor’s task, based on their roles and credentials. In this system, authorization is treated as a text classification problem utilizing fuzzy-neural, self-organizing semantic maps which learn a learn decision criteria based on label information and are capable of generalizing this learned behavior to other information with a zero, or near-zero, false alarm rate. Since “need-to-know” authorizations must be determined for multiple tasks, multiple users, and multiple collections of information, with quick turn-around from request to delivery, the authorization agents must be adaptive and capable of learning new profiles rapidly and with little impact on the overall system performance. We define five different classification methods and provide an architectural framework for the agent system.