2023
Hammad, Sahibzada Saadoon; Iskandaryan, Ditsuhi; Trilles-Oliver, Sergio
An unsupervised TinyML approach applied to the detection of urban noise anomalies under the smart cities environment Journal Article
In: Internet of Things, vol. 23, pp. 100848, 2023, ISSN: 2542-6605.
Abstract | Links | BibTeX | Tags: environmental monitoring, machine learning, TinyML
@article{Saadoon2023a,
title = {An unsupervised TinyML approach applied to the detection of urban noise anomalies under the smart cities environment},
author = {Sahibzada Saadoon Hammad and Ditsuhi Iskandaryan and Sergio Trilles-Oliver},
doi = {10.1016/j.iot.2023.100848},
issn = {2542-6605},
year = {2023},
date = {2023-10-01},
journal = {Internet of Things},
volume = {23},
pages = {100848},
abstract = {Artificial Intelligence of Things (AIoT) is an emerging area of interest, and this can be used to obtain knowledge and take better decisions in the same Internet of Things (IoT) devices. IoT data are prone to anomalies due to various factors such as malfunctioning of sensors, low-cost devices, etc. Following the AIoT paradigm, this work explores anomaly detection in IoT urban noise sensor networks using a Long Short-Term Memory Autoencoder. Two autoencoder models are trained using normal data from two different sensors in the sensor network and tested for the detection of two different types of anomalies, i.e. point anomalies and collective anomalies. The results in terms of accuracy of the two models are 99.99% and 99.34%. The trained model is quantised, converted to TensorFlow Lite format and deployed on the ESP32 microcontroller (MCU). The inference time on the microcontroller is 4 ms for both models, and the power consumption of the MCU is 0.2693 W ± 0.039 and 0.3268 W ± 0.015. Heap memory consumption during the execution of the program for sensors TA120-T246187 and TA120-T246189 is 528 bytes and 744 bytes respectively.},
keywords = {environmental monitoring, machine learning, TinyML},
pubstate = {published},
tppubtype = {article}
}
Artificial Intelligence of Things (AIoT) is an emerging area of interest, and this can be used to obtain knowledge and take better decisions in the same Internet of Things (IoT) devices. IoT data are prone to anomalies due to various factors such as malfunctioning of sensors, low-cost devices, etc. Following the AIoT paradigm, this work explores anomaly detection in IoT urban noise sensor networks using a Long Short-Term Memory Autoencoder. Two autoencoder models are trained using normal data from two different sensors in the sensor network and tested for the detection of two different types of anomalies, i.e. point anomalies and collective anomalies. The results in terms of accuracy of the two models are 99.99% and 99.34%. The trained model is quantised, converted to TensorFlow Lite format and deployed on the ESP32 microcontroller (MCU). The inference time on the microcontroller is 4 ms for both models, and the power consumption of the MCU is 0.2693 W ± 0.039 and 0.3268 W ± 0.015. Heap memory consumption during the execution of the program for sensors TA120-T246187 and TA120-T246189 is 528 bytes and 744 bytes respectively.