Review on Deep Learning Model in Indoor Air Quality Monitoring

Abstract

Maintaining healthy indoor air quality is of primary importance for well-being of individuals, particularly in congested environments such as residences, commercial establishments, and healthcare facilities. Inexpensive, decentralized sensor networks enable continuous and actual management for many IAQ metrics through IoT-related devices. Sensor deficiencies such as drift, bias, noise, and data loss generally undermine the effectiveness of these systems, potentially leading to inaccurate assessments and poor decisions. This research seeks to enhance the efficacy of IoT-based indoor air quality monitoring systems through deep learning techniques for sensor failure mitigation. An extensive review of contemporary research reveals a diverse array of methodologies for addressing sensor faults, encompassing statistical filters, threshold-based approaches, and traditional machine learning algorithms. While these strategies offer partial solutions, they often exhibit insufficient resilience in addressing complex, nonlinear, and time-dependent sensor anomalies characteristic of dynamic indoor environments. In contrast, deep learning models, particularly autoencoders, recurrent neural networks (RNNs), and convolutional neural networks (CNNs), have superior proficiency in learning spatiotemporal patterns, detecting anomalies, and precisely reconstructing corrupted or absent data. The literature study indicates that the incorporation of deep learning frameworks into indoor air quality monitoring systems significantly enhances defect identification and data recovery, hence augmenting system reliability and decision-making precision. Models such as long short-term memory (LSTM) networks and variational autoencoders (VAEs) have shown significant potential in mitigating sensor errors through predictive learning and anomaly correction. The research highlights the potential of deep learning as a transformative approach for developing resilient, intelligent, and fault-tolerant indoor air quality monitoring systems inside the Internet of Things framework.