Telematics Event Modeling for Abnormal Notification Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Telematics systems generate excessive and unmanageable notifications due to continuous data transmission, consuming significant bandwidth in fleet operations and connected vehicles.
Innovation Solution
Implement a telematics server with an event model and notification manager that analyzes telematics data using machine learning to identify unexpected events, reducing notifications to only those related to abnormal conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If continuous notifications are transmitted to keep end users informed, then information completeness is improved, but bandwidth consumption increases and notifications become excessive and unmanageable
Solution Approach 1:
The patent extracts only the essential information (unexpected events) from the continuous stream of telematics data, transmitting only these critical notifications to the notification device while filtering out routine updates, thereby reducing bandwidth consumption while maintaining information completeness for important matters
Solution Approach 2:
Instead of transmitting all telematics data continuously, the system applies partial action by selectively transmitting only a subset of data that represents unexpected events, achieving sufficient information delivery with reduced communication overhead
2Loss of information
If all telematics data is transmitted continuously, then data completeness is improved, but notification manageability deteriorates
Solution Approach 1:
The notification manager extracts and transmits only unexpected events that deviate from normal patterns, filtering out routine operational data, which maintains data completeness for anomaly detection while significantly improving notification manageability by reducing volume
Solution Approach 2:
The system applies different quality levels to different types of data: routine data is filtered out while unexpected events receive full attention and transmission, creating variable notification quality based on local event characteristics rather than uniform transmission
3Measurement precision
If machine learning-based event modeling is implemented, then notification accuracy is improved, but device complexity increases
Solution Approach 1:
The patent introduces an event model as an intermediary layer between raw telematics data and notification generation. This event model, built using machine learning from historical data, acts as a mediator that automatically identifies unexpected events, improving notification accuracy while containing complexity within the modeling component rather than the entire system
Data Source
AI summary
Technologies for reducing event notifications in telematics systems includes a telematics server, a telematics device located in a vehicle, and a notification device. The telematics server receives telematics data from the telematics device and analyzes the telematics data based on an event model to determine whether an unexpected event has occurred. The event model includes a plurality of rules that define expected events related to the vehicle based on historical telematics data. The server generates a notification in response to a determination that the unexpected event has occurred and transmits the notification to the notification device, which may generate a notification to an end user. The event model may be generated and updated by the telematics server using an unsupervised and/or supervised machine learning algorithm based on the telematics data.