NeuralIO: Indoor?Outdoor Detection via Multimodal Sensor Data Fusion on Smartphones

Wang, Long; Sommer, Lennard; Zhou, Yexu; Huang, Yiran; Wang, Jingsi; Riedel, Till; Beigl, Michael

doi:10.18494/sam.2020.2586

Cited by 4 publications

(10 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our work, we compare the performance of different IOD models: ML-based classification models, DL TSC models and threshold-based models. For the ML-based IOD models, the feature variation over time is neglected [15], [24], [25]. We consider two ML architectures that have been used in the past in the context of IOD [15], [17], [24].…”

Section: A Iod Models Consideredmentioning

confidence: 99%

“…For the ML-based IOD models, the feature variation over time is neglected [15], [24], [25]. We consider two ML architectures that have been used in the past in the context of IOD [15], [17], [24]. Specifically, we train (i) an MLP consisting of 5 hidden layers with 1024, 512, 256, 128, and 64 neurons for each hidden layer, respectively, [24], [25] and (ii) a random forest (RF) classifier with 100 trees [15].…”

Section: A Iod Models Consideredmentioning

confidence: 99%

“…We consider two ML architectures that have been used in the past in the context of IOD [15], [17], [24]. Specifically, we train (i) an MLP consisting of 5 hidden layers with 1024, 512, 256, 128, and 64 neurons for each hidden layer, respectively, [24], [25] and (ii) a random forest (RF) classifier with 100 trees [15]. To avoid overffiting in the MLP a dropout layer, with a dropout rate equal to 0.2, is placed after each hidden layer.…”

Section: A Iod Models Consideredmentioning

confidence: 99%

“…However, we note that for the given window value, the difference in the accuracy and the F 1 score between the various DL TSC models is relatively small. We can (i) MLP [24], [25] (ii) RF [15], [25] ( also observe that preprocessing the input time series, i.e., models (iii) and (vi), can lead to a higher accuracy than using raw sensor data (LSTM and ALSTM). For instance, CAP-ALSTM demonstrates a better lower quartile than ALSTM, while DenseNet-LSTM exhibits a better higher top quartile than LSTM.…”

Section: B Comparison Between Different Modelsmentioning

confidence: 99%

See 3 more Smart Citations

Deep-Learning-Based Multivariate Time-Series Classification for Indoor/Outdoor Detection

Bakirtzis

Qiu

Wassell

et al. 2022

IEEE Internet Things J.

View full text Add to dashboard Cite

Recently, the topic of indoor outdoor detection (IOD) has seen its popularity increase, as IOD models can be leveraged to augment the performance of numerous Internet of Things and other applications. IOD aims at distinguishing in an efficient manner whether a user resides in an indoor or an outdoor environment, by inspecting the cellular phone sensor recordings. Legacy IOD models attempt to determine a user's environment by comparing the sensor measurements to some threshold values. However, as we also observe in our experiments, such models exhibit limited scalability, and their accuracy can be poor. Machine learning (ML)-based IOD models aim at removing this limitation, by utilizing a large volume of measurements to train ML algorithms to classify a user's environment. Yet, in most of the existing research, the temporal dimension of the problem is disregarded. In this paper, we propose treating IOD as a multivariate time series classification (TSC) problem, and we explore the performance of various deep learning (DL) models. We demonstrate that a multivariate TSC approach can be used to monitor a user's environment, and predict changes in its state, with greater accuracy compared to conventional approaches that ignore the feature variation over time. Additionally, we introduce a new DL model for multivariate TSC, exploiting the concept of self-attention and atrous spatial pyramid pooling. The proposed DL multivariate TSC framework exploits only low power consumption sensors to infer a user's environment, and it outperforms state-of-the-art models, yielding a higher accuracy combined with a smaller computational cost.

show abstract

Section: A Iod Models Consideredmentioning

confidence: 99%

Section: A Iod Models Consideredmentioning

confidence: 99%

Section: A Iod Models Consideredmentioning

confidence: 99%

Section: B Comparison Between Different Modelsmentioning

confidence: 99%

See 2 more Smart Citations

Deep-Learning-Based Multivariate Time-Series Classification for Indoor/Outdoor Detection

Bakirtzis

Qiu

Wassell

et al. 2022

IEEE Internet Things J.

View full text Add to dashboard Cite

show abstract

“…In the literature, as listed in Table 1, several I/O detection approaches [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28] were proposed based on features extracted from a single sensor and multiple sensors. Many indicators can be extracted from a single sensor, such as GNSS [11,12], WiFi [18,19], cellular [20][21][22], BLE [25], microphone [23], and magnetometer [24]; however, relying solely on one sensor's indicators may degrade the detection robustness under complex scenarios.…”

Section: Handover Mechanism Based On Indoor-outdoor Detectionmentioning

confidence: 99%

SUNS: A User-Friendly Scheme for Seamless and Ubiquitous Navigation Based on an Enhanced Indoor-Outdoor Environmental Awareness Approach

Mansour

Chen

2022

Remote Sensing

View full text Add to dashboard Cite

Ubiquitous and seamless indoor-outdoor (I/O) localization is the primary objective for gaining more user satisfaction and sustaining the prosperity of the location-based services (LBS) market. Regular users, on the other hand, may be unaware of the impact of activating multiple localization sources on localization performance and energy consumption, or may lack experience deciding when to enable or disable localization sources in different environments. Consequently, an automatic handover mechanism that can handle these decisions on a user’s behalf can appreciably improve user satisfaction. This study introduces an enhanced I/O environmental awareness service that provides an automated handover mechanism for seamless navigation based on multi-sensory navigation integration schemes. Moreover, the proposed service utilizes low-power consumption sensor (LPCS) indicators to execute continuous detection tasks and invoke GNSS in confusion scenarios, and transition intervals to make the most firm decision on the credibility of the LPCS-triggered transition and compensate for indicator thresholds. In this manner, GNSS are used for short intervals that help reduce detection latency and power consumption. Consequently, the proposed service guarantees accurate and reliable I/O detection while preserving low power consumption. Leveraging the proposed service as an automated handover helped realize seamless indoor-outdoor localization with less switching latency, using an integrated solution based on extended Kalman filter. Furthermore, the proposed energy-efficient service was utilized to confine crowdsourced data collection to the required areas (indoors and semi-indoors) and prevent excess data collection outdoors, thereby reducing power drainage. Accordingly, the negative impact of data collection on the user’s device can be mitigated, participation can be encouraged, and crowdsourcing systems can be widely adopted.

show abstract

Research trends, themes, and insights on artificial neural networks for smart cities towards SDG-11

Jain

Gue

Jain

2023

Journal of Cleaner Production

View full text Add to dashboard Cite

NeuralIO: Indoor?Outdoor Detection via Multimodal Sensor Data Fusion on Smartphones

Cited by 4 publications

References 24 publications

Deep-Learning-Based Multivariate Time-Series Classification for Indoor/Outdoor Detection

Deep-Learning-Based Multivariate Time-Series Classification for Indoor/Outdoor Detection

SUNS: A User-Friendly Scheme for Seamless and Ubiquitous Navigation Based on an Enhanced Indoor-Outdoor Environmental Awareness Approach

Research trends, themes, and insights on artificial neural networks for smart cities towards SDG-11

Contact Info

Product

Resources

About