A Tensor Completion Approach for Efficient and Robust Fingerprint-based Indoor Localization

TL;DR

This paper introduces a tensor completion method for RF fingerprint-based indoor localization that significantly reduces data collection effort and improves accuracy by robustly handling anomalies in the fingerprint database.

Contribution

It proposes modeling fingerprint data as a tensor and applying tensor decomposition with ADMM for robust recovery, enhancing efficiency and accuracy in indoor localization.

Findings

Achieves 4% error rate with only 10% data sampling

Improves localization accuracy by nearly 20% over existing methods

Reduces data collection effort compared to traditional approaches

Abstract

The localization technology is important for the development of indoor location-based services (LBS). The radio frequency (RF) fingerprint-based localization is one of the most promising approaches. However, it is challenging to apply this localization to real-world environments since it is time-consuming and labor-intensive to construct a fingerprint database as a prior for localization. Another challenge is that the presence of anomaly readings in the fingerprints reduces the localization accuracy. To address these two challenges, we propose an efficient and robust indoor localization approach. First, we model the fingerprint database as a 3-D tensor, which represents the relationships between fingerprints, locations and indices of access points. Second, we introduce a tensor decomposition model for robust fingerprint data recovery, which decomposes a partial observation tensor as the superposition of a low-rank tensor and a spare anomaly tensor. Third, we exploit the alternating direction method of multipliers (ADMM) to solve the convex optimization problem of tensor-nuclear-norm completion for the anomaly case. Finally, we verify the proposed approach on a ground truth data set collected in an office building with size 80m times 20m. Experiment results show that to achieve a same error rate 4%, the sampling rate of our approach is only 10%, while it is 60% for the state-of-the-art approach. Moreover, the proposed approach leads to a more accurate localization (nearly 20%, 0.6m improvement) over the compared approach.