Witness Device Selection via Data Correlation for IoT Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data verification systems in wireless networks, particularly for IoT devices, lack an efficient framework for selecting witness devices based on correlation with primary data generating devices, and fail to adequately address privacy concerns.
Innovation Solution
An apparatus and system that select witness devices by correlating their data with input data from primary devices over a predetermined look-back time period, incorporating privacy provisions to ensure only authorized devices can request and share witness data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional RF eavesdropping is used for witnessing, then the witnessing function is simple to implement, but the verification reliability is insufficient and privacy issues arise
Solution Approach 1:
The patent changes the selection parameter from simple geographic proximity to data correlation degree. Witness devices are selected based on the correlation between their generated data and the primary device's input data, which improves verification reliability by ensuring witnesses actually observed the same event, while the complexity is managed through automated correlation calculations.
Solution Approach 2:
The patent introduces a hub as an intermediary that coordinates between primary devices and witness devices. The hub manages the complex tasks of selecting appropriate witnesses based on data correlation, requesting witness data, and facilitating verification, thereby improving reliability while centralizing the complexity in a dedicated component.
2Ease of operation
If any neighboring device can act as a witness, then the system is easy to operate, but privacy security deteriorates due to unauthorized access
Solution Approach 1:
The patent changes the witness selection criterion from simple co-location to data correlation degree. This ensures that only devices that actually observed the same event can serve as witnesses, automatically preventing unauthorized access while maintaining ease of operation through automated selection based on correlation metrics.
Solution Approach 2:
The system implements feedback mechanisms where the hub evaluates data correlation between potential witnesses and primary devices, and only selects those with sufficient correlation. This feedback loop ensures privacy security by automatically excluding devices that did not observe the actual event, while maintaining operational simplicity through automated evaluation.
3Device complexity
If witness devices are selected based on co-location only, then the selection process is simple, but the measurement precision of verification deteriorates
Solution Approach 1:
The patent changes the selection parameter from geographic location to data correlation degree. This improvement in measurement precision is achieved by selecting witnesses based on how closely their data matches the primary device's data, ensuring they observed the same event. The increased precision is balanced by automated correlation calculations that manage the complexity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention proposes a system and method for recruiting or creating one or more witness devices such that their recent past sensing data has maximum expected correlation to a sensed parameter of importance (e.g., an event's data stream) as recorded by a primary sensor and/or to sub-sets of an overall data stream which are important for models. Thereby, robustness can be increased in situations where a model relies on some unknown combination of data from different sensors, or where there is a risk of eavesdropping by falsely requesting witness data, or where no direct witness device is available and a combination of inputs from several devices is required.