ML Model Predicts Device Capabilities from Usage Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Devices connected to telecommunication networks with unknown technological capabilities pose challenges in predicting the effects of network upgrades and determining infrastructure needs, as their make and model cannot be identified, making it difficult to provide optimal service.
Innovation Solution
A machine learning model is trained using usage data from devices with known capabilities to predict the technological capabilities of devices with unknown capabilities, categorizing them by air interface protocols, voice over LTE support, and LTE band support, enabling better infrastructure planning and service provision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If devices with unknown technological capability are connected to the telecommunication network, then network connectivity and service coverage are improved, but the ability to predict network upgrade effects and plan infrastructure is degraded
Solution Approach 1:
The patent replaces the traditional mechanical approach of identifying devices through make and model databases with a machine learning-based predictive system. The system uses usage data patterns to infer device capabilities, substituting data-driven intelligence for conventional identification methods and resolving the contradiction between serving unknown devices and lacking capability information.
Solution Approach 2:
The patent transforms the approach from using static device identification parameters (make and model) to using dynamic usage data parameters. By analyzing communication patterns, data consumption, and network interaction behaviors, the system infers device capabilities without relying on traditional identification fields, thus maintaining information accuracy while expanding service coverage.
2Measurement precision
If machine learning models are used to predict device capabilities, then infrastructure planning accuracy is improved, but computational resources and data processing requirements are degraded
Solution Approach 1:
The patent creates a universal machine learning model that serves multiple functions: it predicts device capabilities, categorizes devices into groups, and supports infrastructure planning decisions. This multi-functional approach consolidates what would otherwise require separate analytical systems, reducing overall computational requirements while maintaining prediction accuracy across different device types and network scenarios.
Data Source
AI summary
This disclosure describes techniques for determining technological capability of devices whose technological capability are unknown. The devices can be users of a wireless telecommunication network. A device, such as a cell phone, has a type allocation code (TAC) which can indicate a make and model of the device. Once the make and model of the device are known, the technical capability of the device is known. Certain TAC numbers, however, do not indicate the make and model of the device, and thus a device's technical capability is unknown. To determine the technological capability of the devices, a machine learning model is trained using usage data of the devices with known technological capability. After training, the machine learning model can be deployed to predict the technological capability of devices with unrecognizable TAC numbers by providing usage data associated with the devices with unrecognizable TAC numbers.