Wireless Client Throughput Probing for Automatic ML Data Labeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning techniques require large amounts of manually labeled data to determine wireless client health, which is inefficient and difficult to achieve automatically due to the challenges of determining client health in varying wireless conditions.
Innovation Solution
Implement active probing to determine the throughput of wireless clients, using customized burst sizes to accurately capture temporal and spatial variations, and automatically label data for training machine learning algorithms without significant network overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labelling of large-scale data is performed continuously and/or intermittently, then the accuracy of machine learning training data is improved, but the efficiency and feasibility of the labelling process deteriorates
Solution Approach 1:
The system performs self-service by automatically determining client health status through active probing and throughput measurement. The access point autonomously sends probes to mobile devices, measures throughput, and labels data without human intervention, making the labelling process self-sufficient and highly efficient
Solution Approach 2:
The manual mechanical labelling process is replaced with an automated electronic system. Machine learning algorithms and automated probing mechanisms substitute for human operators, using electronic measurements and computational processing to determine health status and assign labels automatically
2Extent of automation
If automatic labelling is implemented without active probing, then the labelling process becomes more automated, but the accuracy of determining client health status deteriorates
Solution Approach 1:
Active probing acts as an intermediary mechanism between the automated labelling system and the actual client health status. The probes serve as a mediator that directly measures throughput and provides objective data about connection quality, enabling accurate automated labelling without human intervention
Solution Approach 2:
The system implements feedback through active probing that continuously measures throughput and provides real-time information about client health status. This feedback loop enables the automated labelling system to accurately determine health status by incorporating direct measurements from the network
3Measurement precision
If active probing with customized burst sizes is performed, then the accuracy of throughput measurement is improved, but the network overhead increases
Solution Approach 1:
The burst size is made dynamic and customized based on specific measurement requirements and network conditions. Rather than using a fixed burst size for all measurements, the system adapts the burst size to capture temporal and spatial variations in throughput, improving measurement accuracy while managing network overhead
Solution Approach 2:
The system applies partial action by using customized burst sizes that are sufficient for accurate measurement without being excessive. The probing is designed to capture necessary throughput variations without overwhelming the network, balancing measurement precision with acceptable overhead
Data Source
AI summary
A method includes sending at least one probe to a mobile device to determine a burst size of at least one burst; sending the at least one burst to the mobile device, the at least one burst includes a first and a second number of probes, the first number of probes are sent to the mobile device at a first time and the second number of probes are sent to the mobile device at a second time after the first time, and the first number of probes and the second number of probes are based on the burst size; receiving an acknowledgement of receipt of the first number of probes at a third time; receiving an acknowledgement of receipt of the second number of probes at a fourth time; and determining, based on a difference between the third time and the fourth time, a throughput of the mobile device.