Remote Sensor Correlation for Distributed Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying events in distributed systems where sensor data availability is limited, making it difficult to locate the source or cause of events and their remote effects.
Innovation Solution
Correlating sensor readings at a location remote from the cause of an event using distributed sensors to identify and control conditions within the system, allowing for the development of models to link events across different locations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor data is collected only at the event location, then measurement precision is improved, but device complexity increases and loss of information occurs in distributed systems
Solution Approach 1:
The patent uses machine learning models as intermediaries to connect sensor data from remote locations with events at the target location. The model learns correlations between remote sensor readings and local events, enabling indirect detection without requiring sensors at every event location.
Solution Approach 2:
The patent transitions from spatial dimension (placing sensors at event locations) to temporal dimension (using historical data for model training). By analyzing temporal patterns in remote sensor data, the system can detect events without physical proximity.
2Loss of information
If distributed sensors are deployed across multiple locations, then loss of information is reduced, but device complexity increases
Solution Approach 1:
The patent creates a universal machine learning model that can process sensor data from multiple different locations and sensor types. The model is trained on diverse data and can detect various events across the distributed system using a single analytical framework.
Solution Approach 2:
The patent extracts the essential correlation patterns from training data and embeds them in a machine learning model. This extracted knowledge allows the system to detect events using simplified processing of remote sensor data without requiring complex real-time analysis at each sensor location.
3Measurement precision
If machine learning models are trained with labeled data from event locations, then measurement precision is improved, but loss of time occurs during data collection
Solution Approach 1:
The patent performs preliminary action by collecting and labeling training data in advance during normal system operation. The machine learning model is trained beforehand with labeled examples, so when events occur, the system can quickly identify them without requiring real-time data collection and labeling.
Solution Approach 2:
The system uses normal operational sensor data for training purposes. During regular system operation, sensor readings are automatically collected and labeled when events occur, creating training data without requiring separate data collection campaigns or system shutdowns.
Data Source
AI summary
A method of identifying parameters associated with an event comprises identifying an event at a first location, correlating the event with one or more sensor outputs, identifying at least one sensor output of the one or more sensor outputs correlated with the event at the first location; and displaying the at least one sensor output along with an indication of the event. The one or more sensor outputs are obtained from a location other than the first location.