Table 1 Comparative analysis of various existing research in the field of Air-Quality Analysis.

Ahmed et al.¹

CNN-RNN model with satellite-derived hydro-climatological variables

Satellite-derived hydro-climatological data

Effective in areas with sparse ground-based sensors; improved AQI forecasting

High computational demand; challenges in real-time scalability

Rabie et al.²

CNN-BiLSTM hybrid framework

Urban air quality monitoring datasets

Firm spatial resolution; effective in megacities for localized predictions

High training complexity; computational resource dependency

Kumar & Kumar³

Multi-view Stacked CNN-BiLSTM (MvS CNN-BiLSTM)

Meteorological and emissions data from Indian cities

Robust to varying conditions; integrates multiple data views for comprehensive modeling

Resource-intensive; lacks sensor-agnostic capabilities

Wu et al.⁵

DVMD Informer-CNN-LSTM optimised with Dung Beetle Algorithm

Historical AQI datasets

Improved noise handling and nonlinear data modeling through DVMD; optimized hyperparameters

High computational cost due to the optimization algorithm

Barthwal & Goel⁴

DCNN and LSTM architectures

AQI time-series data from urban India

Strong temporal dependency modeling is effective in urban settings

Scalability Issues: challenging to deploy in real-time applications

Prado-Rujas & García-Dopico⁶

Sensor-agnostic deep learning framework

Spatio-temporal air quality data

Effective in diverse environments, sensor-agnostic design increases adaptability

Limited focus on high computational efficiency for large-scale deployments

Liu et al.⁷

Machine learning for air quality class prediction

Monitoring station data

Simplicity and efficiency in classifying air quality

Struggles with capturing complex Spatio-temporal dependencies

Wang et al.⁸

Graph Convolution and Multi-Head Attention Mechanism

Urban pollution datasets

Improved Spatio-temporal modeling; advanced attention mechanism for feature selection

Computationally expensive; difficult to interpret results