Rolling Baseline Engine for Anomalous Subject Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing surveillance technologies fail to effectively identify anomalous subjects and devices in crowded sites using a rolling baseline, particularly in environments with various sensors and personal devices, and lack methods for monitoring valuable assets at manufacturing facilities.
Innovation Solution
A system that deploys sensors using wired or wireless technologies to gather data from subjects and personal devices, analyzes it using a rolling baseline engine that establishes a conceptual hypercube with a centroid, scoring incoming packets to identify anomalies based on clustering techniques like k-means, and utilizes wireless antennas for better location determination and data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional surveillance technologies are used in crowded sites, then basic monitoring is achieved, but the ability to identify anomalous subjects and devices effectively is insufficient
Solution Approach 1:
The system segments the surveillance problem into multiple dimensions by creating a hypercube model with different axes representing various behavioral and contextual parameters. Each dimension of the hypercube captures specific aspects of subject behavior, allowing the system to analyze anomalies in a structured, multi-faceted manner rather than as a single complex pattern recognition problem.
Solution Approach 2:
The patent transforms the anomaly detection problem from traditional 2D or 3D analysis into a high-dimensional hypercube space. By mapping subject behaviors, device interactions, temporal patterns, and contextual factors into multiple dimensions, the system achieves superior anomaly detection capability while maintaining manageable complexity through dimensional organization.
2Reliability
If a rolling baseline with hypercube clustering is implemented, then anomaly detection accuracy improves, but computational requirements and processing complexity increase
Solution Approach 1:
The system performs preliminary actions by continuously maintaining a rolling baseline of normal behavior patterns in the hypercube space before anomalies occur. This pre-computed baseline allows the system to quickly compare incoming data against established norms, reducing the computational burden during real-time anomaly detection compared to computing patterns from scratch each time.
Solution Approach 2:
The rolling baseline mechanism is self-updating and self-maintaining, automatically adapting to changing normal behaviors without requiring external intervention or reconfiguration. The system serves itself by continuously refining its own baseline model based on incoming data, reducing the need for manual tuning and external computational resources.
3Adaptability or versatility
If multiple sensors and personal devices are deployed, then data coverage and monitoring capability improve, but difficulty in processing and analyzing data from various sources increases
Solution Approach 1:
The hypercube model serves as a universal framework that can accommodate multiple types of sensors and data sources through its dimensional structure. Different sensor inputs (video, audio, RFID, environmental sensors) and device data can be mapped to appropriate dimensions or combinations of dimensions, allowing the same analytical engine to process diverse data types without requiring separate processing pipelines for each source.
Data Source
AI summary
Techniques are disclosed for identifying anomalous subjects and devices at a site. The devices may or may not be carried by or associated with subjects at the site. A number of various types of sensors may be utilized for this purpose. The sensors gather data about the subjects and devices. The data is processed by a data processing module which provides its output to a rolling baseline engine. The rolling baseline engine establishes a baseline for what is considered the “normal” behavior for subjects/devices at the site based on a desired dimension of analysis. Data associated with subjects/devices that is not normal is identified as an anomaly along with the associated subject/device. The findings are archived for performing analytics as required.