What is it about?

FedHIL is a practical solution for indoor localization. It is designed to ensure that your smartphone accurately and reliably determines your exact position within complex indoor environments, such as shopping malls, airports, or hospitals. What sets FedHIL apart is its remarkable ability to maintain accuracy using an innovative distributed machine learning approach, even when dealing with different types of smartphones or when faced with the challenges of complex indoor environments. It's like having a built-in GPS for indoor spaces. But here's what truly makes FedHIL exceptional: it places a premium on safeguarding your privacy. With FedHIL, the user’s personal data remains completely confidential and secure. Extensively tested in real-world scenarios, FedHIL has consistently outperformed existing methods. On average, it boosts your smartphone's localization accuracy by 1.62 times. This unique combination of privacy protection, high accuracy, and resilience in complex indoor settings makes FedHIL a practical choice for indoor localization.

Featured Image

Why is it important?

FedHIL's novel contributions are pivotal for indoor localization. Its novel data augmentation minimizes signal discrepancies between devices, resulting in highly accurate localization. The integration of federated learning serves multiple crucial purposes, including ensuring robust user-privacy by facilitating collaborative model training without sharing sensitive location data directly. FedHIL also introduces a novel aggregation method that efficiently combines data while remaining resilient to dynamic noise from the environment and heterogeneous devices. Lastly, we propose a lightweight design that supports efficient deployment on resource-constrained devices such as smartphones, enhancing accessibility. FedHIL maintains its accuracy and reliability in challenging environments and diverse data sources, offering a robust, privacy-conscious, and smartphone-friendly indoor localization solution.


FedHIL effectively addresses several critical concerns: the demand for precise localization accuracy in dense urban environments, the challenge of maintaining accuracy consistently indoors, the necessity for lightweight operation on resource-constrained devices, and the paramount importance of safeguarding user privacy. The integration of federated learning is particularly noteworthy, offering a user-friendly means of collaborative model creation without compromising data security. Moreover, FedHIL's lightweight design positions it as a practical solution for diverse users, potentially reshaping the way we navigate through indoor spaces while ensuring user privacy.

Danish Gufran
Colorado State University

Read the Original

This page is a summary of: FedHIL: Heterogeneity Resilient Federated Learning for Robust Indoor Localization with Mobile Devices, ACM Transactions on Embedded Computing Systems, September 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3607919.
You can read the full text:



The following have contributed to this page