WiFi-Based Human Identification via Convex Tensor Shapelet Learning

• DOI: 10.1609/aaai.v32i1.11497
• Publication Date: 2022-06-24T21:33:21Z
• Authors: Han Zou; Yuxun Zhou; Jianfei Yang; Weixi Gu; Lihua Xie; Costas Spanos
• Keywords: Identification; Implementation; Regularization

We propose AutoID, a human identification system that leverages the measurements from existing WiFi-enabled Internet of Things (IoT) devices and produces identity estimation via novel sparse representation learning technique. The key idea is to use unique fine-grained gait patterns each person revealed WiFi Channel State Information (CSI) measurements, technically referred as shapelet signatures, "fingerprint" for identification. For this purpose, OpenWrt-based IoT platform designed collect CSI data commercial devices. More importantly, we new optimization-based framework tensors, namely Convex Clustered Concurrent Shapelet Learning (C3SL), which formulates problem convex optimization. global solution C3SL can be obtained efficiently with generalized gradient-based algorithm, three concurrent regularization terms reveal inter-dependence clustering effect tensor data. Extensive experiments are conducted in multiple real-world indoor environments, showing AutoID achieves an average accuracy 91% group 20 people. As combination sensing platform, attains substantial progress towards more accurate, cost-effective sustainable pervasive implementations.