Robust ultra-wideband range error mitigation with deep learning at the edge

Angarano, Simone; Mazzia, Vittorio; Salvetti, Francesco; Fantin, Giovanni; Chiaberge, Marcello

doi:10.1016/j.engappai.2021.104278

Cited by 42 publications

(20 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The original design of the Range Error Mitigation Network (REMNet) presented in [1] is adapted to be executed in real-time on a low-power microcontroller. The neural network should be fully quantized to perform all the operations with 8-bit integers and meet the real-time constraints.…”

Section: Network Designmentioning

confidence: 99%

Ultra-Low-Power Range Error Mitigation for Ultra-Wideband Precise Localization

Angarano

Salvetti

Mazzia

et al. 2022

Lecture Notes in Networks and Systems

Self Cite

View full text Add to dashboard Cite

Precise and accurate localization in outdoor and indoor environments is a challenging problem that currently constitutes a significant limitation for several practical applications. Ultra-wideband (UWB) localization technology represents a valuable low-cost solution to the problem. However, non-line-of-sight (NLOS) conditions and complexity of the specific radio environment can easily introduce a positive bias in the ranging measurement, resulting in highly inaccurate and unsatisfactory position estimation. In the light of this, we leverage the latest advancement in deep neural network optimization techniques and their implementation on ultra-low-power microcontrollers to introduce an effective range error mitigation solution that provides corrections in either NLOS or LOS conditions with a few mW of power. Our extensive experimentation endorses the advantages and improvements of our low-cost and power-efficient methodology.

show abstract

Section: Network Designmentioning

confidence: 99%

Ultra-Low-Power Range Error Mitigation for Ultra-Wideband Precise Localization

Angarano

Salvetti

Mazzia

et al. 2022

Lecture Notes in Networks and Systems

Self Cite

View full text Add to dashboard Cite

show abstract

“…ANN based method in VLP (visible light positioning) [18] NLOS classification and mitigation based on RSSI [20] and TOA [58] DNN based device-free localization [21] CNN and DNN completion and refinement for EDM recovery [28] Hybrid SVM-and DNN-based method [28] KNN and Naive Bayes methods with RSSI fingerprints [24] CNN-LSTM-based hybrid deep learning with RSSI heat map [39] SVM and Gaussian Process regressions for LOS/NLOS identification, classification and error mitigation [101][102][103][104][105][106][107][108] ANN and CNN based method to identify and to estimate position of room with human object [109] Unsupervised ML Isloation forest-based classification method [19] Ranging module-based NN method for trilateration [22] k-means RSSI-based classification for improving accuracy [25] VAE-based semi-supervised learning model with latent variables [38] PDR-based reliable unsupervised approach with iBeacon corrections and fingerprint database auto-building [13] Other contributions such as [38] work on the integration of NN-based techniques to avoid the limitation because of the non-Gaussian inverse problem. To realize the coordinates' prediction with WIFI fingerprints, which is a sequential time-series regression problem, the authors of [38] proposed to add CNN in order to capture the features and to learn the Gaussian density of the complex high-dimensional input, before analyzing the state transition time-series data of hidden layers by RNN.…”

Section: Supervised MLmentioning

confidence: 99%

“…Last but not least, a new development trend has been revealed in recent papers [104,106], which analyze the effect of multi-path using machine-learning-based localization methods, especially related to NLOS identification and classification issue [101][102][103][104][105][106][107][108]. The authors of [104] studied the multi-path conditions between two transceivers, by analyzing the impact of obstacles on the prediction localization error.…”

Section: Supervised MLmentioning

confidence: 99%

A Survey of Recent Indoor Localization Scenarios and Methodologies

Tian

Cabani

Chafouk

2021

Sensors

View full text Add to dashboard Cite

Recently, various novel scenarios have been studied for indoor localization. The trilateration is known as a classic theoretical model of geometric-based indoor localization, with uniform RSSI data that can be transferred directly into distance ranges. Then, a trilateration solution can be algebraically acquired from theses ranges, in order to fix user’s actual location. However, the collected RSSI or other measurement data should be further processed and classified to lower the localization error rate, instead of using the raw data influenced by multi-path effect, multiple nonlinear interference and noises. In this survey, a large number of existing techniques are presented for different indoor network structures and channel conditions, divided as LOS (light-of-sight) and NLOS (non light-of-sight). Besides, the input measurement data such as RSSI (received signal strength indication), TDOA (time difference of arrival), DOA (distance of arrival), and RTT (round trip time) are studied towards different application scenarios. The key localization techniques like RSSI-based fingerprinting technique are presented using supervised machine learning methods, namely SVM (support vector machine), KNN (K nearest neighbors) and NN (neural network) methods, especially in an offline training phase. Other unsupervised methods as isolation forest, k-means, and expectation maximization methods are utilized to further improve the localization accuracy in online testing phase. For Bayesian filtering methods, apart from the basic linear Kalman filter (LKF) methods, nonlinear stochastic filters such as extended KF, cubature KF, unscented KF and particle filters are introduced. These nonlinear methods are more suitable for dynamic localization models. In addition to the localization accuracy, the other important performance features and evaluation aspects are presented in our paper: scalability, stability, reliability, and the complexity of proposed algorithms is compared in this survey. Our paper provides a comprehensive perspective to compare the existing techniques and related practical localization models, with the aim of improving localization accuracy and reducing the complexity of the system.

show abstract

“…The remaining 3141 instances are used as validation to find the most promising hyperparameters with a grid search analysis. All results, training, and testing code for the AcT model are open source and publicly available 1 .…”

Section: A Experimental Settingsmentioning

confidence: 99%

“…The benchmark is executed for both OpenPose and PoseNet data and repeated for all the three train/test splits provided by MPOSE2021. The baselines chosen for the benchmark are a Multilayer Perceptron (MLP), a fully convolutional model (Conv1D), and REMNet, which is a more sophisticated convolutional network with attention and residual blocks proposed in [1] for time series feature extraction. Moreover, four popular models used for multivariate time series classification and, in particular, HAR are reproduced and tested.…”

Section: B Action Recognition On Mpose2021mentioning

confidence: 99%

Action Transformer: A Self-Attention Model for Short-Time Pose-Based Human Action Recognition

Mazzia,

Angarano,

Salvetti

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Deep neural networks based purely on attention have been successful across several domains, relying on minimal architectural priors from the designer. In Human Action Recognition (HAR), attention mechanisms have been primarily adopted on top of standard convolutional or recurrent layers, improving the overall generalization capability. In this work, we introduce Action Transformer (AcT), a simple, fully selfattentional architecture that consistently outperforms more elaborated networks that mix convolutional, recurrent, and attentive layers. In order to limit computational and energy requests, building on previous human action recognition research, the proposed approach exploits 2D pose representations over small temporal windows, providing a low latency solution for accurate and effective real-time performance. Moreover, we open-source MPOSE2021, a new large-scale dataset, as an attempt to build a formal training and evaluation benchmark for real-time short-time human action recognition. Extensive experimentation on MPOSE2021 with our proposed methodology and several previous architectural solutions proves the effectiveness of the AcT model and poses the base for future work on HAR.

show abstract

Robust ultra-wideband range error mitigation with deep learning at the edge

Cited by 42 publications

References 29 publications

Ultra-Low-Power Range Error Mitigation for Ultra-Wideband Precise Localization

Ultra-Low-Power Range Error Mitigation for Ultra-Wideband Precise Localization

A Survey of Recent Indoor Localization Scenarios and Methodologies

Action Transformer: A Self-Attention Model for Short-Time Pose-Based Human Action Recognition

Contact Info

Product

Resources

About