TY - GEN
T1 - Indoor Localization and Obstacle Detection using Spatio-Temporal Autoencoders
AU - Savvidis, Lazaros S.
AU - Mavromoustakis, Constandinos X.
AU - Markakis, Evangelos K.
AU - Mastorakis, George
AU - Bourdena, Athina
AU - Batalla, Jordi Mongay
N1 - Publisher Copyright:
© 2025 IEEE.
PY - 2025
Y1 - 2025
N2 - Accurate and robust indoor localization is vital for Internet of Things (IoT) location-based services, such as navigation and asset tracking. Dynamic indoor environments with moving objects or new obstacles, can degrade a localization system's performance, unless it adapts to changes. This paper introduces a Spatio Temporal Autoencoder (STAE) module to detect environmental changes, by monitoring deviations in Time of Flight (ToF) measurements. The STAE learns normal spatio-temporal patterns from Fine Time Measurement (FTM) data, and flags anomalies when observations deviate significantly from the baseline. This unsupervised data-driven module, enhances the FTM and Pedestrian Dead Reckoning (PDR) framework from our prior work, improving its resilience to dynamic conditions. Simulations show that the STAE reliably identifies nodes that become unreachable after an obstacle appears. By integrating the STAE into our system, we maintain accuracy in changing indoor environments.
AB - Accurate and robust indoor localization is vital for Internet of Things (IoT) location-based services, such as navigation and asset tracking. Dynamic indoor environments with moving objects or new obstacles, can degrade a localization system's performance, unless it adapts to changes. This paper introduces a Spatio Temporal Autoencoder (STAE) module to detect environmental changes, by monitoring deviations in Time of Flight (ToF) measurements. The STAE learns normal spatio-temporal patterns from Fine Time Measurement (FTM) data, and flags anomalies when observations deviate significantly from the baseline. This unsupervised data-driven module, enhances the FTM and Pedestrian Dead Reckoning (PDR) framework from our prior work, improving its resilience to dynamic conditions. Simulations show that the STAE reliably identifies nodes that become unreachable after an obstacle appears. By integrating the STAE into our system, we maintain accuracy in changing indoor environments.
KW - fine time measurement (FTM)
KW - indoor localization
KW - pedestrian dead reckoning (PDR)
KW - spatio-temporal autoencoder (STAE)
KW - Wi-Fi Aware
UR - https://www.scopus.com/pages/publications/105033352364
U2 - 10.1109/NFV-SDN66355.2025.11349575
DO - 10.1109/NFV-SDN66355.2025.11349575
M3 - Conference contribution
AN - SCOPUS:105033352364
T3 - 2025 IEEE Conference on Network Function Virtualization and Software-Defined Networking, NFV-SDN 2025
BT - 2025 IEEE Conference on Network Function Virtualization and Software-Defined Networking, NFV-SDN 2025
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2025 IEEE Conference on Network Function Virtualization and Software-Defined Networking, NFV-SDN 2025
Y2 - 10 November 2025 through 12 November 2025
ER -