[Paper Review] CRISLoc: Reconstructable CSI Fingerprintingfor Indoor Smartphone Localization

Zhihui Gao, Yunfan Gao+3 more|arXiv (Cornell University)|Oct 15, 2019

Indoor and Outdoor Localization Technologies47 references4 citations

TL;DR

CRISLoc is a passive, smartphone-based indoor localization system that reconstructs CSI fingerprints using transfer learning and joint clustering/outlier detection to maintain accuracy despite access point (AP) changes. It achieves a mean localization error of 0.29m in a 6m×8m lab, with only 5.4cm and 8.6cm increases when one or two APs are moved, demonstrating robustness and real-world deployability on off-the-shelf smartphones without AP cooperation.

ABSTRACT

Channel state information (CSI) based fingerprinting for WIFI indoor localization has attracted lots of attention very recently.The frequency diverse and temporally stable CSI better represents the location dependent channel characteristics than the coarsereceived signal strength (RSS). However, the acquisition of CSI requires the cooperation of access points (APs) and involves only dataframes, which imposes restrictions on real-world deployment. In this paper, we present CRISLoc, the first CSI fingerprinting basedlocalization prototype system using ubiquitous smartphones. CRISLoc operates in a completely passive mode, overhearing thepackets on-the-fly for his own CSI acquisition. The smartphone CSI is sanitized via calibrating the distortion enforced by WiFi amplifiercircuits. CRISLoc tackles the challenge of altered APs with a joint clustering and outlier detection method to find them. A novel transferlearning approach is proposed to reconstruct the high-dimensional CSI fingerprint database on the basis of the outdated fingerprintsand a few fresh measurements, and an enhanced KNN approach is proposed to pinpoint the location of a smartphone. Our studyreveals important properties about the stability and sensitivity of smartphone CSI that has not been reported previously. Experimentalresults show that CRISLoc can achieve a mean error of around 0.29m in a6m times 8mresearch laboratory. The mean error increases by 5.4 cm and 8.6 cm upon the movement of one and two APs, which validates the robustness of CRISLoc against environment changes.

Motivation & Objective

• To enable CSI-based fingerprinting localization on off-the-shelf smartphones without requiring AP cooperation or connection.
• To address the challenge of dynamic AP deployments by detecting altered APs and reconstructing outdated CSI fingerprints.
• To improve localization accuracy and robustness in real-world indoor environments using calibrated, high-dimensional CSI from smartphones.
• To develop a practical, deployable system that overcomes limitations of existing CSI toolkits requiring active connections or specialized hardware.

Proposed method

• Uses Nexmon toolkit to extract raw CSI from data frames, ACKs, and beacons on smartphones in passive mode, without establishing connections to APs.
• Applies AGC calibration and subcarrier filtering to stabilize CSI amplitudes, enabling use of ~50 subcarriers—nearly double that of Intel 5300 CSI Tool.
• Employs a joint clustering and outlier detection method to identify altered APs by analyzing spatial clustering patterns of estimated positions.
• Proposes a novel transfer learning framework that projects outdated and fresh CSI into a shared subspace to reconstruct obsolete fingerprints.
• Introduces an enhanced KNN (EEKNN) matching rule that incorporates adaptive weights based on variance and spatial correlation to improve localization accuracy.
• Uses sequential analysis and Jenks optimization to set reliability thresholds for detecting AP changes and validating fingerprint consistency.

Experimental results

Research questions

• RQ1 Can CSI fingerprinting be effectively implemented on off-the-shelf smartphones in a completely passive mode without AP cooperation?
• RQ2 How can altered APs be detected automatically when their positions or identities change in a dynamic indoor environment?
• RQ3 To what extent can outdated CSI fingerprints be reconstructed using transfer learning to maintain localization accuracy after AP changes?
• RQ4 What is the impact of AP movement on localization error, and can the system remain robust under such environmental changes?

Key findings

• CRISLoc achieves a mean localization error of 0.29m in a 6m×8m laboratory environment using only smartphone-based CSI acquisition.
• The system maintains high accuracy with only a 5.4cm increase in mean error when one AP is moved, and 8.6cm when two APs are moved, demonstrating strong robustness.
• The joint clustering and outlier detection method achieves F1-scores of 0.893 and 0.827 for detecting altered APs when one and two APs are changed, respectively.
• Transfer learning enables effective reconstruction of outdated CSI fingerprints, preserving system accuracy without full re-surveying.
• The EEKNN matching rule reduces localization error by 21.3% compared to traditional WKNN by incorporating adaptive variance-based weighting.
• The system successfully extracts usable CSI from Nexus 5 smartphones using 20MHz bandwidth in 2.4GHz band without establishing connections to APs.