<ul><li>Researchers have proposed a permutation-invariant neural architecture for indoor localization using RSSI scans from Wi-Fi access points.</li><li>The model processes each scan as an unordered set of (BSSID, RSSI) pairs, mapping BSSIDs to learned embeddings and incorporating signal strength.</li><li>Utilizing a Set Transformer, the model deals with variable-length, sparse inputs and learns attention-based representations over access point relationships.</li><li>Experimental results on a campus dataset show that the proposed model accurately captures spatial structure and performs competitively, particularly excelling in scenarios involving multiple buildings and floors.</li></ul>

Permutation-Invariant Transformer Neural Architectures for Set-Based Indoor Localization Using Learned RSSI Embeddings

Discover more