Project Category

Indoor Localization & Representation Learning

Research methods for estimating indoor position from signal fingerprints, with an emphasis on candidate selection, representation learning, and efficient inference.

All projects
Machine Learning Deep Learning Localization Signal Processing WiFi 5G

GMM-TSFA Indoor Localization

GitHub

Task: estimate indoor position from WiFi fingerprints.

Method: fingerprint transformation, similarity filtering, and adaptive reference selection for explainable matching.

Machine Learning WiFi Representation Learning Similarity Filtering

Group Matching Method

GitHub

Task: reduce the candidate search space in RSS fingerprint matching.

Method: group-based matching that studies localization accuracy alongside computational efficiency.

Machine Learning Search-Space Reduction Efficient Inference

PPSA-Net Structured Attention

Task: model relationships among signal features for indoor localization.

Method: prior-probability-driven structured attention for signal co-occurrence patterns.

Deep Learning Structured Attention Co-occurrence Modeling

LSTM Distill MLP

GitHub

Task: connect temporal signal modeling with lightweight localization inference.

Method: a dual-mode framework combining LSTM modeling and MLP knowledge distillation.

Deep Learning LSTM Model Distillation Efficient Inference