Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MG-HGNN: A Heterogeneous GNN Framework for Indoor Wi-Fi Fingerprint-Based Localization

Yibu Wang, Zhaoxin Zhang, Ning Li, Xinlong Zhao, Dong Zhao, Tianzi Zhao

TL;DR

This work tackles indoor Wi-Fi fingerprint localization by addressing limitations of single-graph and homogeneous GNN approaches for RSSI data. It introduces MG-HGNN, a framework with two graph-construction branches (position-based and feature-based) and a heterogeneous GNN that fuses multi-graph information for accurate localization. Key contributions include a task-directed multi-graph construction strategy, a modular heterogeneous GNN design, and an optional online adapter to enhance generalization in online settings. Experiments on UJIIndoorLoc and UTSIndoorLoc demonstrate superior mean location error and competitive floor/building classification, validating the framework as a practical design paradigm for GNN-based localization.

Abstract

Received signal strength indicator (RSSI) is the primary representation of Wi-Fi fingerprints and serves as a crucial tool for indoor localization. However, existing RSSI-based positioning methods often suffer from reduced accuracy due to environmental complexity and challenges in processing multi-source information. To address these issues, we propose a novel multi-graph heterogeneous GNN framework (MG-HGNN) to enhance spatial awareness and improve positioning performance. In this framework, two graph construction branches perform node and edge embedding, respectively, to generate informative graphs. Subsequently, a heterogeneous graph neural network is employed for graph representation learning, enabling accurate positioning. The MG-HGNN framework introduces the following key innovations: 1) multi-type task-directed graph construction that combines label estimation and feature encoding for richer graph information; 2) a heterogeneous GNN structure that enhances the performance of conventional GNN models. Evaluations on the UJIIndoorLoc and UTSIndoorLoc public datasets demonstrate that MG-HGNN not only achieves superior performance compared to several state-of-the-art methods, but also provides a novel perspective for enhancing GNN-based localization methods. Ablation studies further confirm the rationality and effectiveness of the proposed framework.

MG-HGNN: A Heterogeneous GNN Framework for Indoor Wi-Fi Fingerprint-Based Localization

TL;DR

This work tackles indoor Wi-Fi fingerprint localization by addressing limitations of single-graph and homogeneous GNN approaches for RSSI data. It introduces MG-HGNN, a framework with two graph-construction branches (position-based and feature-based) and a heterogeneous GNN that fuses multi-graph information for accurate localization. Key contributions include a task-directed multi-graph construction strategy, a modular heterogeneous GNN design, and an optional online adapter to enhance generalization in online settings. Experiments on UJIIndoorLoc and UTSIndoorLoc demonstrate superior mean location error and competitive floor/building classification, validating the framework as a practical design paradigm for GNN-based localization.

Abstract

Received signal strength indicator (RSSI) is the primary representation of Wi-Fi fingerprints and serves as a crucial tool for indoor localization. However, existing RSSI-based positioning methods often suffer from reduced accuracy due to environmental complexity and challenges in processing multi-source information. To address these issues, we propose a novel multi-graph heterogeneous GNN framework (MG-HGNN) to enhance spatial awareness and improve positioning performance. In this framework, two graph construction branches perform node and edge embedding, respectively, to generate informative graphs. Subsequently, a heterogeneous graph neural network is employed for graph representation learning, enabling accurate positioning. The MG-HGNN framework introduces the following key innovations: 1) multi-type task-directed graph construction that combines label estimation and feature encoding for richer graph information; 2) a heterogeneous GNN structure that enhances the performance of conventional GNN models. Evaluations on the UJIIndoorLoc and UTSIndoorLoc public datasets demonstrate that MG-HGNN not only achieves superior performance compared to several state-of-the-art methods, but also provides a novel perspective for enhancing GNN-based localization methods. Ablation studies further confirm the rationality and effectiveness of the proposed framework.

Paper Structure

This paper contains 41 sections, 20 equations, 11 figures, 11 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional ML Based Approaches
  4. Deep Learning Based Localization
  5. Graph Neural Network Based Fingerprinting Methods
  6. Framework Overview
  7. Branch 1: Position-based Graph Construction
  8. Branch 2: Feature-based Graph Construction
  9. Branch 3: Heterogeneous GNN based Localization
  10. Offline and Online Work Flow
  11. Multi-Graph Heterogeneous GNN Framework for Fingerprint-Based Localization
  12. Data Preprocessing
  13. Universal Graph Construction - Branch 1
  14. Sampling Point Aggregation
  15. KNN Search
  16. ...and 26 more sections

Figures (11)

  • Figure 1: Conceptual framework of heterogeneous GNN based fingerprint localization.
  • Figure 2: MG-HGNN Framework overview. Branches 1 and 2 are the graph construction branches, and branch 3 is the localization branch.
  • Figure 3: Sampling points of UJIIndoorLoc dataset.
  • Figure 4: Validation graph of UJIIndoorLoc Building 0.
  • Figure 5: Structure of GNN-based localization.
  • ...and 6 more figures