Semi-Supervised Learning-Enhanced Fingerprint Indoor Positioning by Exploiting an Adapted Mean Teacher Model

Peng Chen,Yingzhi Liu,Wei Li,Jingyi Wang,Jianxiu Wang,Bei Yang,Gang Feng

DOI: https://doi.org/10.3390/electronics13020298

IF: 2.9

2024-01-10

Electronics

Abstract:Location awareness is crucial for numerous emerging wireless indoor applications. Deep learning algorithms have demonstrated the potential for achieving the required level of positioning accuracy in indoor environments. However, obtaining abundant labels for data-driven machine learning is costly in practical situations. As an effective solution to alleviating the insufficiency of labeled data for deep learning-based indoor positioning, deep semi-supervised learning (DSSL) can be employed to lessen the dependency on labeled data by exploiting potential patterns in unlabeled samples. In this paper, we propose an Adapted Mean Teacher (AMT) model within the DSSL paradigm for indoor fingerprint positioning by using a channel impulse response. To enhance the generalization of the trained model, we design an efficient implicit augmentation scheme for the training process in the AMT model. Furthermore, we develop a tailored residual network to efficiently extract location characteristics in the AMT framework. We conduct extensive simulation experiments for indoor scenarios with heavy non-line-of-sight conditions based on open datasets to demonstrate the effectiveness of our proposed AMT model. Numerical results indicate that the AMT model outperforms several consistency regularization methods and the pseudo-label method in terms of positioning accuracy and lower positioning latency, achieving a mean error of 90cm when using a small number of labels.

engineering, electrical & electronic,computer science, information systems,physics, applied

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve high - precision positioning in an indoor environment. In particular, under heavy non - line - of - sight (NLOS) conditions, it uses channel impulse response (CIR) data to improve the accuracy of indoor fingerprint positioning. Traditional methods face the problem of high cost and difficulty in obtaining a large amount of labeled data in practical applications, which limits the application effect of deep - learning models in indoor positioning. For this reason, the paper proposes an improved mean - teacher model (Adapted Mean Teacher, AMT) based on deep semi - supervised learning (DSSL), aiming to reduce the dependence on labeled data and enhance the generalization ability of the model by using the potential patterns in unlabeled samples. Specifically, the paper designs an efficient implicit data - augmentation scheme and develops a customized residual network to effectively extract location features, thereby achieving sub - meter - level positioning accuracy with a small number of labels.