WiFi Indoor Localization via Unsupervised Domain Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current WiFi-based localization systems face challenges in accurately localizing users with different body shapes and in varying environments, as WiFi signals are sensitive to human body shapes and environmental changes, leading to inconsistent location fingerprints and reduced accuracy.
Innovation Solution
The Widora system employs a domain-adaptive approach with a data augmenter using Variational Autoencoders (VAEs) to generate synthetic WiFi fingerprints and a joint classification-reconstruction neural network to adapt to new users and environments without requiring explicit labeling, utilizing unlabelled data to improve localization accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If WiFi fingerprint-based localization is used, then localization capability is provided, but accuracy deteriorates due to sensitivity to body shapes and environmental changes
Solution Approach 1:
The system performs preliminary domain adaptation training using unlabeled data from new users before actual localization. The neural network model is pre-adjusted to the new user's characteristics through unsupervised domain adaptation, extracting domain-invariant features that are robust to body shape and environmental variations, thereby improving localization accuracy for new users without requiring labeled data
Solution Approach 2:
The system changes the parameter representation by transforming WiFi fingerprint data into domain-invariant features through neural network processing. The domain adaptation process modifies feature extraction parameters to focus on user-specific characteristics while filtering out environmental variations, enabling accurate localization despite sensitivity to environmental changes
2Measurement precision
If labeled data from multiple users is collected to improve accuracy, then localization performance improves, but system complexity and data collection requirements increase
Solution Approach 1:
The system enables self-service localization for new users by automatically performing domain adaptation using unlabeled WiFi fingerprint data. The neural network model autonomously adjusts to new users without requiring manual data collection, labeling, or system reconfiguration, eliminating the complexity of multi-user data collection while maintaining high localization accuracy
Solution Approach 2:
The system performs preliminary unsupervised domain adaptation training using only unlabeled data from new users before localization tasks. This preliminary action pre-adjusts the model to new users' characteristics without requiring labeled data or complex data collection processes, simplifying the system while improving accuracy for new users
Data Source
Figure 1A~1B
Figure 2
Figure 3
AI summary
A location-aware electronic device is provided. The electronic device trains feature extraction layers, reconstruction layers, and classification layers. The training may be based on a reconstruction loss and/or a clustering loss. The electronic device processes a fingerprint to obtain an augmented fingerprint using randomization based on statistics of the fingerprint. The feature extraction layers provide feature data to both the reconstruction layers and the classification layers. The classification layers operate on the codes to obtain an estimated location label. An application processor operates on the estimated location label to provide a location-aware application result to a person.