Self-adaptive WiFi Localization via Crowd-sourced Signal Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing WiFi indoor localization methods are inefficient in providing accurate region-level location awareness in dynamic indoor environments, as they fail to adapt to changes in Wi-Fi distributions and require continuous user data contributions for model updates.
Innovation Solution
A self-adaptive device employing machine learning algorithms, specifically one-vs-all support vector machines and Platt scaling techniques, that uses crowd-sourced Wi-Fi signal strength data to identify regions and adapt models in real-time, enabling reliable indoor location awareness and self-diagnosis in dynamic Wi-Fi environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional WiFi localization methods are used, then device complexity is reduced, but measurement precision and adaptability deteriorate due to inability to adapt to dynamic WiFi distributions
Solution Approach 1:
The system performs self-service through automatic model adaptation using crowd-sourced data. The self-adaptive device automatically detects changes in WiFi distributions and updates localization models without requiring manual intervention, thereby maintaining high measurement precision while managing complexity through automation
Solution Approach 2:
The system implements feedback mechanisms by continuously collecting crowd-sourced WiFi signal strength data and using it to update localization models. This feedback loop enables the system to adapt to dynamic WiFi environments, improving measurement precision while the automated nature of the feedback process helps manage device complexity
2Adaptability or versatility
If continuous model updates with user data are implemented, then adaptability improves, but loss of time and productivity decrease due to required user contributions
Solution Approach 1:
The system merges multiple data sources and functions into a unified localization service. By combining crowd-sourced data from multiple users and integrating detection, classification, and model update functions into a single system, the patent achieves high adaptability while reducing the time burden on individual users through data aggregation
Solution Approach 2:
The system implements multi-functionality by serving multiple purposes: localization, model training, and environmental adaptation all within a single platform. This universal approach allows the system to gather diverse data for model updates without requiring dedicated time contributions for each function, thereby improving adaptability while minimizing time loss
3Reliability
If self-adaptive mechanisms are added, then reliability improves through automatic adaptation, but device complexity increases due to additional algorithms and data processing
Solution Approach 1:
The system replaces complex mechanical or manual adjustment mechanisms with automated algorithmic processes. By using machine learning algorithms to automatically detect and adapt to WiFi distribution changes, the system achieves high reliability in dynamic environments while the substitution of automated processes for manual ones helps manage the complexity burden
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A robust WiFi indoor localization system for large public sites includes a self-adaptive system communicatively coupled to a plurality of clients. The self-adaptive system includes a processor, a computer readable medium, and a communication interface module.