A hierarchical auxiliary deep neural network architecture for large-scale indoor localization based on Wi-Fi fingerprinting

Cha, Jaehoon; Lim, Enggee

doi:10.1016/j.asoc.2022.108624

Cited by 22 publications

(13 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this section, to evaluate the influence of differences in datasets from different months on the robustness of the model in a dynamic environment, we use each month's datasets as a validation of the VITL algorithmand compare it with the conventional machine learning approach and the deep learning approach. We used five conventional machine learning algorithms, SVM [22], KNN [23], RF [24], DT [25], and GNB [26], for comparison, after which they showed the best performance with those in [44] and [45] a baseline neural network comprising two fully connected hidden layers, with 128 and 68 nodes, and five deep learning algorithms CNN [27], C-FNN1, HADNN1 [46] and rrifloc [47] were compared. In figure 5, the DT algorithm shows poor robustness with other conventional machine learning algorithms, such as the GNB algorithm, after the ninth month, although the other conventional machine learning algorithms are slightly better but still start to float more in the ninth month.…”

Section: Algorithmic Comparison Of Vtil and Conventional Machine Lear...mentioning

confidence: 99%

VTIL: A multi-layer indoor location algorithm for RSSI images based on vision transformer

Zhou,

Yang,

Deng

et al. 2024

Eng. Res. Express

View full text Add to dashboard Cite

With the popularization and application of WiFi technology, precise location in multi-layer indoor environments play an increasingly important role in current public places. However, there are still problems such as noise and multi-path effects caused by the propagation degradation of the Received Signal Strength Indicator (RSSI) in indoor scenarios and the shading of obstacles, which lead to a reduction in indoor location accuracy. In this paper, we propose a Vision Transformer indoor localization (VTIL) algorithm based on RSSI and Vision-Transformer (ViT). First, the RSSI fingerprint is normalized to the maximum and minimum values, and the Principal Component Analysis (PCA) method is used for feature extraction to reduce the influence of noise or unimportant features. Secondly, the RSSI fingerprint library is converted into an RSSI gray image library. Then the RSSI gray image is divided into several small blocks, and the position-coding input is performed in the form of a sequence. The ViT model divides the weight ratio of each block in the RSSI image to alleviate the impact of multi-path effects. According to the experimental results using public datasets, our method reduces the average distance estimation error by 37.26% compared with the currently known indoor fingerprint localization methods.

show abstract

Section: Algorithmic Comparison Of Vtil and Conventional Machine Lear...mentioning

confidence: 99%

VTIL: A multi-layer indoor location algorithm for RSSI images based on vision transformer

Zhou,

Yang,

Deng

et al. 2024

Eng. Res. Express

View full text Add to dashboard Cite

show abstract

“…In 2022, Wang et al designed a novel DNN-based indoor positioning framework called CHISEL, which combines a convolutional encoder and a CNN classifier [ 37 ]. Jaehoon Cha et al propose a hierarchical auxiliary deep neural network called HADNN, which uses a continuous feedforward network to identify buildings and estimate the floor coordinates [ 38 ].…”

Section: Related Workmentioning

confidence: 99%

“…To further evaluate the performance of our improved DAE and enhanced DNN algorithms, we compare them with state-of-the-art deep learning indoor positioning algorithms, including SAEDNN [ 34 ], CNNLoc [ 35 ], CCpos [ 36 ], CHISEL [ 37 ], CHISEL-DA [ 37 ], and HADNN [ 38 ]. The results are shown in Table 4 and Figure 8 , respectively.…”

Section: Performance Evaluationmentioning

confidence: 99%

“…It can be seen from Table 5 that the average positioning error of machine learning methods in specific-location positioning is 6.13 m at the lowest (KNN) and 10.42 m at the highest (DT and GNB), while the DAE model proposed in this paper reduces to 6.01 m. Similarly, the state-of-the-art deep learning algorithms used for comparison include SAEDNN [ 34 ], CNNLoc [ 35 ], CCpos [ 36 ], CHISEL [ 37 ], CHISEL-DA [ 37 ], and HADNN [ 38 ]. Figure 9 shows the comparison results of DDAE and the advanced deep learning algorithms.…”

Section: Performance Evaluationmentioning

confidence: 99%

See 1 more Smart Citation

Calibration-Free 3D Indoor Positioning Algorithms Based on DNN and DIFF

Yang

Deng

et al. 2022

Sensors

View full text Add to dashboard Cite

The heterogeneity of wireless receiving devices, co-channel interference, and multi-path effect make the received signal strength indication (RSSI) of Wi-Fi fluctuate greatly, which seriously degrades the RSSI-based positioning accuracy. Signal strength difference (DIFF), a calibration-free solution for handling the received signal strength variance between diverse devices, can effectively reduce the negative impact of signal fluctuation. However, DIFF also leads to the explosion of the RSSI data dimension, expanding the number of dimensions from m to Cm2, which reduces the positioning efficiency. To this end, we design a data hierarchical processing strategy based on a building-floor-specific location, which effectively improves the efficiency of high-dimensional data processing. Moreover, based on a deep neural network (DNN), we design three different positioning algorithms for multi-building, multi-floor, and specific-location respectively, extending the indoor positioning from the single plane to three dimensions. Specifically, in the stage of data preprocessing, we first create the original RSSI database. Next, we create the optimized RSSI database by identifying and deleting the unavailable data in the RSSI database. Finally, we perform DIFF processing on the optimized RSSI database to create the DIFF database. In the stage of positioning, firstly, we design an improved multi-building positioning algorithm based on a denoising autoencoder (DAE). Secondly, we design an enhanced DNN for multi-floor positioning. Finally, the newly deep denoising autoencoder (DDAE) used for specific location positioning is proposed. The experimental results show that the proposed algorithms have better positioning efficiency and accuracy compared with the traditional machine learning algorithms and the current advanced deep learning algorithms.

show abstract

“…Moreover, KNN relies on stationary signal information and does not handle temporal dynamics well, further limiting its effectiveness. In contrast, deep neural networks (DNNs) can address the issues in large-scale indoor localization [13], [14], which can model complex relationships between the input features and output labels and thereby efficiently handle the spatial variability and dynamic signals encountered in large-scale multi-building and multi-floor indoor localization. In addition to classical feedforward neural networks (FNNs) used in earlier works (e.g., [13]), more advanced DNNs like convolutional neural networks (CNNs) [15], [16] and recurrent neural networks (RNNs) [17] are employed as well due to their improved robustness and generalization capability.…”

Section: Introductionmentioning

confidence: 99%

On the Use and Construction of Wi-Fi Fingerprint Databases for Large-Scale Multi-Building and Multi-Floor Indoor Localization: A Case Study of the UJIIndoorLoc Database

Li,

Tang,

Kim

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

A hierarchical auxiliary deep neural network architecture for large-scale indoor localization based on Wi-Fi fingerprinting

Cited by 22 publications

References 14 publications

VTIL: A multi-layer indoor location algorithm for RSSI images based on vision transformer

VTIL: A multi-layer indoor location algorithm for RSSI images based on vision transformer

Calibration-Free 3D Indoor Positioning Algorithms Based on DNN and DIFF

On the Use and Construction of Wi-Fi Fingerprint Databases for Large-Scale Multi-Building and Multi-Floor Indoor Localization: A Case Study of the UJIIndoorLoc Database

Contact Info

Product

Resources

About