Advanced Fusion Technologies Based on Heterogeneous Data for Indoor Positioning

Description of the special session:

We are delighted to announce a special session on "Advanced Fusion Technologies Based on Heterogeneous Data for Indoor Positioning" at the upcoming 14th International Conference on Indoor Positioning and Indoor Navigation (IPIN 2024). This session aims to explore the cutting-edge methodologies and applications in the realm of indoor positioning and navigation systems. At this stage, heterogeneous data fusion is proved to be an effective approach for accuracy enhancement of indoor positioning under complex and large-scaled environments, which combines advantages of heterogeneous location and sensing sources – such as RF signals, inertial sensors, magnetic fields, acoustic signals, and visual information – presents a promising avenue for enhancing the accuracy, reliability, and robustness of indoor positioning systems. In this special session, we focus on providing a research and discussion platform for advanced heterogeneous data fusion structures in indoor positioning field, including but not limited to KF/PF, graph optimization, deep-learning, and other merging fusion technologies. Submissions should present novel findings, innovative ideas, or state-of-the-art reviews that have not been published elsewhere. We encourage interdisciplinary approaches that combine insights from computer science, electrical engineering, geoinformatics, and other relevant fields. Accepted papers will be presented during the special session and will be included in the conference proceedings. This is an excellent opportunity to share your work with the international community and to network with peers who share your interest in advancing indoor positioning technologies. Let us work together to promote the continuous development of heterogeneous data fusion technologies and strive to achieve more ubiquitous, more accurate, and more universal indoor navigation. 

Novelty and motivation:

Heterogeneous data fusion stands as a pivotal strategy for enhancing the performance and stability of indoor positioning systems, surpassing the capabilities of single-source solutions. Despite its potential, the integration of multi-source fusion architectures in both academic and practical domains encounters significant challenges. These include:

  1. The need for effective methodologies to assess time-varying errors and to determine the appropriate fusion weights for disparate positioning sources within the fusion process.
  2. The imperative to judiciously select positioning sources, ensuring that the accuracy and computational efficiency of the multi-source fusion are not compromised.
  3. The requirement to account for environmental and contextual influences, with the aim of refining the outcomes of the multi-source fusion.

Addressing these three critical issues is essential for realizing stable and efficient multi- source fusion. They represent pressing and complex hurdles that the field of indoor positioning must urgently overcome. Distinguished from the regular session, this special session zeroes in on the enhancement of precision within the existing heterogeneous data fusion architectures, as well as the exploration of nascent data fusion technologies tailored for indoor positioning. Our research ambit, while broad, can be tentatively segmented into four key dimensions, though our interests are not confined to these alone:

  1. Filtering technologies based on classic or enhanced KF and PF;
  2. Graph optimization technologies based on graph/ factor graph optimization et al;
  3. Data-driven technologies based on supervised/unsupervised deep learning;
  4. Multi-driven technologies based on a mixture of different fusion methods.
  5. Heterogeneous data sources can be adopted for fusion-based indoor positioning, eg: Wi-Fi, inertial sensors, UWB, BLE, Visual, Lidar, BIM. 

Scope of the session:

In this Special Session, we invite authors to submit papers related (but not limited) to:

  • Heterogeneous data sources for fusion-based positioning
  • Classical or enhanced KF and PF based fusion methods
  • Graph optimization or factor graph optimization
  • Supervised/unsupervised deep learning
  • Data-driven or data augmentation methods
  • Pre-training or fine-turning models
  • Large models including large language models, etc.
  • Multimodal and multi-task learning models 


  • Heterogeneous data sources
  • Fusion-based Positioning
  • KF/PF
  • Graph optimization
  • Deep-learning
  • Data-driven
  • Data augmentation
  • Pre-training/ fine-turning
  • Large models
  • Multimodal/multi-task

Session chairs:

  • Dr. Yue Yu (The Hong Kong Polytechnic University)
  • Dr. Lei Wang (Wuhan University)