CRISLoc is the first CSI fingerprinting based localization prototype system using ubiquitous smartphones.

To the best of our knowledge, CRISLoc is the first CSI fingerprinting localization system using off-the-shelf smartphones.

Router and Nexus 5

Extensive experiments show that CRISLoc achieves a mean error of 0.29m in a research laboratory (6m×8m), and a median error of 0.78m in a complex academic building (8m×28m) consisting of a research room, a long orridor and a small square.

Floor Plan for Reseach Laboratory
Floor Plan for Academic Building

CSI is extracted from Nexus 5 by Nexmon, whose amplitudes are more stable over time.

Heatmap of CSI’s 49 subcarriers over time by Nexmon

We design a suit of methods to sanitize CSI data, encompassing the cancellation of automatic gain control and the filtering of unstable subcarriers and frames.

abnormal CSI

We design a joint clustering and outlier detection approach to find the altered APs. The core idea of this approach is to perform localization by different subsets of APs.

Subsets localization
Altered AP detection in academic building

We develop a novel transfer learning approach to reconstruct their CSI & RSSI fingerprints.

CSI and RSSI localization when a)none alters, b)un-reconstructed, and c)reconstructed in academic building

Our data is now available on Google Drive, please refer to the link below:

To reference CRISLoc:

Zhihui Gao*, Yunfan Gao*, Sulei Wang, Dan Li, Yuedong Xu, “CRISLoc: Reconstructable CSI Fingerprinting for Indoor Smartphone Localization”, submitted to IEEE Internet of Things Journal.