ULA-MSRTI: The Unsupervised Link Analysis Based Movement Superposition Radio Tomographic Imaging in Passive UHF RFID Scenarios

Abstract

Device-free localization (DFL) is a promising technique for localizing targets without any auxiliary devices. Passive ultra high frequency radio frequency identification (pUHF RFID) based radio tomographic imaging (RTI) emerges as a particularly favorable DFL approach due to its inherent advantages. However, accurately localizing multiple targets remains a challenge for it. To address the challenge, we propose a novel unsupervised link analysis based movement superposition RTI (ULA-MSRTI), including movement superposition RTI (MSRTI) model and the unsupervised link analysis (ULA) scheme. MSRTI models the influence of target movement on wireless signals at every moment. Then, we design the ULA scheme to determine received signal strength (RSS) variations of target-affected links required for MSRTI. Firstly, the scheme adopts statistical analysis to remove wireless links that are mainly affected by noise. Then, we design an adaptive selection algorithm based on density-based spatial clustering of applications with noise to eliminate wireless links affected by burst signals and self-environmental multipath. After that, the scheme matches RSS variations of remaining links with the single-target database to recover RSS variations of target-affected links at one moment. Finally, we input recovered RSS variations into MSRTI to obtain the locations of multiple targets. The experiment shows that ULA-MSRTI's performance improves by an average of over 14.6% compared to other methods, and ULA-MSRTI is the insensitive to the locations of multiple targets.