AI Asset Control Classification via Telemetry and Tagging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Versatile field assets, such as computing devices and equipment, face challenges in maintaining interoperability among friendly users while preventing theft, spoofing, or compromise by unfriendly actors.
Innovation Solution
A system that uses computing devices to monitor telemetry data from in-field assets, including position, orientation, and biometric data, and employs artificial intelligence models to classify possession or control of target assets, distinguishing between friendly and unfriendly actors, and adjusting trust levels accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If field assets are made interoperable among friendly users, then versatility and ease of operation are improved, but the risk of theft, spoofing, or compromise by unfriendly actors increases
Solution Approach 1:
The patent introduces a server system as an intermediary that receives telemetry data from field assets and tagging data from tagging assets. This mediator processes the data and determines possession or control status, thereby enabling interoperability while filtering out unfriendly access attempts. The server acts as a trusted third party that coordinates asset sharing without exposing assets directly to potential threats.
Solution Approach 2:
The system continuously monitors telemetry data from field assets and tagging data from tagging assets, creating a feedback loop that provides real-time information about asset possession and control status. This feedback mechanism allows the system to dynamically adjust trust levels and prevent unauthorized access, maintaining security while enabling versatile interoperability among friendly users.
2Reliability
If monitoring and classification systems are implemented to prevent unauthorized access, then security and reliability are improved, but device complexity and difficulty of operation increase
Solution Approach 1:
The patent divides the security system into distinct functional components: field assets that provide telemetry data, tagging assets that provide possession information, a server system that processes and analyzes data, and AI models that perform classification. This segmentation allows each component to be independently developed, deployed, and maintained, reducing overall system complexity while maintaining high reliability through distributed functionality.
Solution Approach 2:
The system enables field assets and tagging assets to automatically provide their own data (telemetry and tagging data respectively) without requiring manual intervention. The AI models automatically analyze the data and determine possession status, and the system automatically adjusts trust levels. This self-service approach reduces operational complexity while maintaining robust security through automated monitoring and classification.
3Measurement precision
If AI models are trained with extensive telemetry and tagging data to accurately classify asset possession, then measurement precision and reliability are improved, but loss of time during training and data collection increases
Solution Approach 1:
The patent performs AI model training in advance during a training phase, separating the time-consuming training process from the operational phase. During training, the system collects and processes telemetry and tagging data to create trained AI models. Once trained, these models can quickly and accurately classify asset possession during run-time operations without requiring extensive data collection time, thus achieving high measurement precision with minimal operational time loss.
Data Source
AI summary
Examples are disclosed that relate to methods and systems for classifying the possession or control of a target asset. One example provides a system comprising one or more computing devices having processors and associated memories storing instructions executable by the processors. The instructions are executable to conduct a simulation or observation of an in-field event comprising a plurality of actors controlling a plurality of in-field assets. The system is further configured to monitor telemetry data from each of the in-field assets. In addition, tagging data is received from an in-field asset under the control of a member of the friendly group. A training data set is generated including the telemetry data and the tagging data, and an artificial intelligence model is trained to predict whether a run-time target asset is in the possession or control of the friendly or the unfriendly actor, or lost, based on run-time telemetry data.