Mobile Terminal Predictive Model Update via Threshold-Based Feed
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current predictive modeling techniques in mobile telecommunications face challenges in updating predictive models efficiently, as they require frequent data exchanges that consume precious radio bandwidth and are not suitable for mobile networks.
Innovation Solution
A method for updating predictive models on mobile terminals, where the terminal measures and estimates variables, generates a feed message if the difference between measured and estimated values exceeds a threshold, and sends it to an update server for model updates, minimizing data exchanges and only replacing the model if the contribution is significant, with the update server controlling the frequency of updates based on the feed threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If predictive models are updated regularly to integrate changes in the environment, then prediction accuracy is improved, but radio bandwidth consumption increases
Solution Approach 1:
The system changes the parameter of data exchange frequency based on the computed difference metric. When the difference between measured and estimated values exceeds a threshold, model updates are triggered; otherwise, updates are postponed. This dynamic parameter adjustment optimizes the trade-off between prediction accuracy and bandwidth consumption.
Solution Approach 2:
The system implements a feedback mechanism where the mobile terminal computes the difference between measured and estimated values, compares it to a threshold, and only initiates model updates when the difference is significant. This feedback loop ensures updates occur only when necessary, reducing unnecessary bandwidth consumption while maintaining prediction accuracy.
2Reliability
If learning data is periodically captured and transferred to the learning device, then predictive model relevance is improved, but network data flow overhead increases
Solution Approach 1:
The system extracts only the essential information needed for model updating by computing and transmitting only the difference between measured and estimated values when it exceeds a threshold. This selective extraction reduces the quantity of data transmitted over the network while ensuring model relevance is maintained.
Solution Approach 2:
The system dynamically changes the parameter of data transmission by adjusting the frequency and volume of learning data transfers based on the computed difference metric. Updates are performed only when the difference exceeds a threshold, thereby reducing network data volume while maintaining model reliability.
3Adaptability or versatility
If predictive models are updated frequently, then adaptation to environmental changes is improved, but computational overhead increases
Solution Approach 1:
The system uses a feedback mechanism where the computed difference between measured and estimated values is compared to a threshold to determine when model updates are necessary. This feedback control ensures the model adapts to environmental changes when needed while avoiding unnecessary computational overhead from frequent updates.
Solution Approach 2:
The system dynamically adjusts the update frequency parameter based on the computed difference metric. When the difference exceeds the threshold, the system triggers a model update to adapt to environmental changes; otherwise, it maintains the current model, thereby reducing computational overhead while preserving adaptability.
Data Source
AI summary
A method for updating a predictive model of a variable representing the operation of a mobile terminal connected to a communication network by packets is described. A first predictive model is configured to estimate a value of the variable as a function of the value of predictors, linked to the variable by a common operating context. The method can be implemented by the mobile terminal and can include generating a feed message, comprising at least the measured value and the values of the predictors, if the difference between the measured value and the estimated value is greater than or equal to a determined threshold, called the feed threshold, the method can further include transmitting the at least one feed message generated, to an update server connected to the network, receiving an update message, comprising a second predictive model updated on the basis of at least the feed message, coming from the update server, and replacing the first predictive model with the second predictive model.