IoT Behavior Monitoring via Multi-Dimensional Device Grouping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for monitoring large sets of IoT devices are resource-intensive, requiring numerous assessment modules to identify abnormal behavior, which is inefficient and costly.
Innovation Solution
A multi-dimensional monitoring system assigns IoT devices to grid points in an orthonormal space, using a subset of monitors per dimension to detect deviations from behavior models, and combines outputs to identify non-compliant devices, employing one-class classification and machine learning on frame headers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If one uses a conventional monitoring system with one assessment module per IoT device, then each device can be monitored individually for abnormal behavior, but the resource consumption becomes extremely high when dealing with large sets of devices
Solution Approach 1:
The patent segments the monitoring task by grouping IoT devices into clusters based on their behavior models. Instead of monitoring each device individually with a separate assessment module, devices are organized into groups where a single assessment module can evaluate multiple devices simultaneously. This segmentation approach reduces the total number of assessment modules needed while maintaining effective monitoring coverage.
Solution Approach 2:
The patent implements multi-functionality by enabling a single assessment module to serve multiple IoT devices within a cluster. The assessment module evaluates behavior models for multiple devices concurrently, making it a universal monitor that performs multiple monitoring functions without requiring dedicated resources for each device. This significantly reduces resource consumption while maintaining monitoring capability.
2Adaptability or versatility
If one increases the number of assessment modules to monitor more IoT devices, then coverage increases, but the system complexity and cost increase proportionally
Solution Approach 1:
The patent divides the large set of IoT devices into manageable clusters or groups based on similarity in behavior models. Each cluster is monitored by a dedicated assessment module, creating a hierarchical segmentation structure. This allows the system to scale adaptably - new devices can be added to existing clusters or form new clusters without proportionally increasing overall system complexity, as each cluster shares monitoring resources.
Solution Approach 2:
The patent merges multiple monitoring functions into fewer assessment modules by grouping devices with similar behavior characteristics. Instead of having separate assessment modules for each device, devices are combined into clusters that share common monitoring infrastructure. This merging approach increases the number of monitorable devices while reducing system complexity through resource sharing and functional consolidation.
3Measurement precision
If one uses machine learning models to assess device behavior, then detection accuracy improves, but the computational requirements and resource usage increase
Solution Approach 1:
The patent segments the computational load by organizing devices into clusters and assigning dedicated assessment modules to each cluster. Instead of running comprehensive machine learning analyses on all devices simultaneously, the system performs targeted ML assessments within smaller clusters. This segmentation distributes computational requirements across multiple smaller tasks rather than one large computationally intensive operation, reducing overall power requirements while maintaining detection accuracy.
Solution Approach 2:
The patent applies partial action by using machine learning models selectively - not for every possible device parameter, but for the most critical behavior indicators that distinguish normal from abnormal operation. The assessment modules focus ML analysis on key behavior models and parameters rather than exhaustively analyzing all possible device characteristics. This partial approach maintains sufficient detection accuracy while significantly reducing computational power requirements.
Data Source
AI summary
Methods and apparatus automatically identify which Internet-of-Things (IoT) devices within a set are behaving in a manner non-compliant with a target behavior. Each IoT device is assigned to a grid point in a notional m-dimensional space. A respective assessment module is arranged to monitor behavior of a group of IoT devices assigned to grid points aligned with one another at a respective position along the respective dimension, and to produce an output indicative of non-compliant behavior if the monitoring indicates that behavior in the group of IoT devices deviates from a behavioral model of the IoT devices of said group. An identification module identifies at least one non-compliant IoT device in the set of IoT devices by combining outputs from the assessment modules assigned to the different dimensions of the space. The assessment modules may use trained machine-learning algorithms embodying a model of normal behavior of the group of devices.