Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LocaGen: Low-Overhead Indoor Localization Through Spatial Augmentation

Abdelrahman Abdelmotlb, Abdallah Taman, Sherif Mostafa, Moustafa Youssef

TL;DR

LocaGen tackles the high data collection burden of indoor fingerprinting by introducing a diffusion-model-based spatial augmentation that can generate realistic fingerprints at unseen locations using data from a subset of seen locations. It combines a density-driven seen/unseen location division, domain-specific augmentation of seen data, and a spatially aware conditional diffusion model to synthesize high-quality fingerprints for unseen coordinates. The approach maintains localization accuracy with significantly reduced surveying and demonstrates up to 28% improvement over state-of-the-art augmentation methods on a real WiFi RSS dataset. Practically, LocaGen enables plug-in deployment in fingerprinting-based localization systems, reducing overhead without sacrificing performance, and offers a scalable path toward deployable indoor localization solutions.

Abstract

Indoor localization systems commonly rely on fingerprinting, which requires extensive survey efforts to obtain location-tagged signal data, limiting their real-world deployability. Recent approaches that attempt to reduce this overhead either suffer from low representation ability, mode collapse issues, or require the effort of collecting data at all target locations. We present LocaGen, a novel spatial augmentation framework that significantly reduces fingerprinting overhead by generating high-quality synthetic data at completely unseen locations. LocaGen leverages a conditional diffusion model guided by a novel spatially aware optimization strategy to synthesize realistic fingerprints at unseen locations using only a subset of seen locations. To further improve our diffusion model performance, LocaGen augments seen location data based on domain-specific heuristics and strategically selects the seen and unseen locations using a novel density-based approach that ensures robust coverage. Our extensive evaluation on a real-world WiFi fingerprinting dataset shows that LocaGen maintains the same localization accuracy even with 30% of the locations unseen and achieves up to 28% improvement in accuracy over state-of-the-art augmentation methods.

LocaGen: Low-Overhead Indoor Localization Through Spatial Augmentation

TL;DR

LocaGen tackles the high data collection burden of indoor fingerprinting by introducing a diffusion-model-based spatial augmentation that can generate realistic fingerprints at unseen locations using data from a subset of seen locations. It combines a density-driven seen/unseen location division, domain-specific augmentation of seen data, and a spatially aware conditional diffusion model to synthesize high-quality fingerprints for unseen coordinates. The approach maintains localization accuracy with significantly reduced surveying and demonstrates up to 28% improvement over state-of-the-art augmentation methods on a real WiFi RSS dataset. Practically, LocaGen enables plug-in deployment in fingerprinting-based localization systems, reducing overhead without sacrificing performance, and offers a scalable path toward deployable indoor localization solutions.

Abstract

Indoor localization systems commonly rely on fingerprinting, which requires extensive survey efforts to obtain location-tagged signal data, limiting their real-world deployability. Recent approaches that attempt to reduce this overhead either suffer from low representation ability, mode collapse issues, or require the effort of collecting data at all target locations. We present LocaGen, a novel spatial augmentation framework that significantly reduces fingerprinting overhead by generating high-quality synthetic data at completely unseen locations. LocaGen leverages a conditional diffusion model guided by a novel spatially aware optimization strategy to synthesize realistic fingerprints at unseen locations using only a subset of seen locations. To further improve our diffusion model performance, LocaGen augments seen location data based on domain-specific heuristics and strategically selects the seen and unseen locations using a novel density-based approach that ensures robust coverage. Our extensive evaluation on a real-world WiFi fingerprinting dataset shows that LocaGen maintains the same localization accuracy even with 30% of the locations unseen and achieves up to 28% improvement in accuracy over state-of-the-art augmentation methods.

Paper Structure

This paper contains 18 sections, 1 equation, 6 figures.

Table of Contents

  1. Introduction
  2. The LocaGen Framework
  3. Framework Overview
  4. Design Challenges and Framework Details
  5. Optimizing Seen Locations Selection
  6. The Risk of Generative Model Overfitting
  7. Our Generative Model
  8. Model Design
  9. Integrating Spatial Awareness
  10. Framework Evaluation
  11. Dataset and Experimental Setup
  12. Ablation Study
  13. Verifying our Fingerprint Initializer
  14. Verifying our Spatially-Aware Optimization
  15. Overall Framework Performance
  16. ...and 3 more sections

Figures (6)

  • Figure 1: LocaGen Framework Architecture.
  • Figure 2: LocaGen's spatial augmentation diffusion-based model. The model is optimized with our novel spatially-aware loss function $\mathcal{L}(\theta)$ to ensure synthesizing high-quality samples for the target unseen locations.
  • Figure 3: Impact of our Fingerprint Initializer.
  • Figure 4: Impact of our spatially-aware optimization.
  • Figure 5: Effect of the number of seen locations.
  • ...and 1 more figures