Wireless Flow Adjustment for Rare Resource Demand Samples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing supervised learning models for resource demand prediction in wireless networks face challenges with data insufficiency and bias, leading to poor generalization and inaccurate predictions, especially for rare or underrepresented application scenarios.
Innovation Solution
A system and method to adjust flow characteristics in existing data flows to create new samples that represent underrepresented scenarios, using supervised learning to estimate resource demand, without introducing artificial traffic, by temporarily overriding QoS requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised learning models use existing data for resource demand prediction, then the model can be trained with available data, but the prediction accuracy for rare or underrepresented scenarios remains poor due to data insufficiency and bias
Solution Approach 1:
The system performs preliminary analysis of existing data to identify underrepresented scenarios before training the model. By proactively detecting data gaps and generating synthetic samples for rare scenarios in advance, the system ensures better data coverage and improves prediction accuracy for previously underrepresented cases without waiting for natural data accumulation
Solution Approach 2:
The system creates synthetic data samples by copying and adapting existing data patterns to represent underrepresented scenarios. By generating artificial training samples that mimic the characteristics of rare scenarios, the system enriches the training dataset and enables the model to learn from scenarios that would otherwise have insufficient real-world data
2Quantity of substance
If the system creates synthetic samples by adjusting flow characteristics, then data representation for rare scenarios is improved, but the complexity of data processing and sample generation increases
Solution Approach 1:
The system changes parameters of existing data flows (such as traffic volume, timing patterns, or service types) to generate synthetic samples representing underrepresented scenarios. By systematically varying flow parameters within realistic bounds, the system creates diverse training samples without requiring complex data generation architectures or external data sources
Solution Approach 2:
The system uses its own existing data infrastructure and flow characteristics to generate synthetic samples, rather than relying on external data sources or complex third-party tools. By leveraging internally available data and processing capabilities, the system reduces external dependencies and simplifies the overall data generation process
3Ease of manufacture
If existing data is used for training, then the training process is simple, but the model fails to capture relevant relations for rare scenarios due to high bias in the data
Solution Approach 1:
The system introduces an intermediary data layer that bridges existing training data and rare scenarios. By using synthetic samples as an intermediate representation, the system preserves the simplicity of training on available data while gradually incorporating patterns from underrepresented scenarios, thus maintaining training ease while reducing information loss
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
A system (1, 21) for obtaining samples for a dynamic resource management system in a wireless network is configured to analyze (101) data for a dynamic resource management system to determine one or more combinations of input values which are not present or underrepresented in the data. The dynamic resource management system uses supervised learning to estimate resource demand. The system is further configured to adjust (103) one or more flow characteristics of a data flow between a user device and a base station to obtain one of the combinations of input values, determine a quantity of resources used for the data flow and actual state and/or cell characteristics relevant to the data flow, create a new sample based on the used quantity of resources, the adjusted flow characteristics, and the actual state and/or cell characteristics, and store the new sample in the data.