MTFALoc: A Novel CSI-based Multi-target Device-free Passive Fingerprinting Localization via Multi-Label Classification

In recent years, location-based service applications such as cashier-less shopping, mobile advertisement targeting, and geo-based augmented reality (AR) have significantly increased. These applications rely on indoor localization technology to provide users with convenient and interactive experiences based on their current locations. Despite advancements, existing research primarily focuses on individual target localization, often overlooking the challenges of localizing multiple targets simultaneously. In this paper, we propose MTFALoc, a novel device-free passive fingerprinting localization system that uses CSI and multi-label classification to address this gap. MTFALoc transforms multi-target localization into a multi-label classification problem by dividing the area into training point regions and using advanced preprocessing to extract high-resolution FMTF from CSI. Leveraging spatio-temporal redundancy in amplitude and phase information, MTFALoc enhances localization accuracy. It also features a Transformer-based network with a pixel attention model for precise feature localization. Experimental results show that MTFALoc outperforms existing systems, offering significant improvements in accuracy and robustness, making it a promising advancement in indoor multi-target localization.