Multi-Source Fusion Enhanced Power-Efficient Sustainable Computing for Air Quality Monitoring

Cai, J, Liu, T, Wang, T, Feng, H, Fang, K, Bashir, AK ORCID: https://orcid.org/0000-0001-7595-2522 and Wang, W (2024) Multi-Source Fusion Enhanced Power-Efficient Sustainable Computing for Air Quality Monitoring. IEEE Internet of Things Journal. p. 1.

Preview

Accepted Version Available under License In Copyright. Download (16MB) | Preview

Abstract

Given the severity of air pollution, air quality monitoring has become a crucial aspect of Artificial Intelligence of Things (AIoT) applications, providing essential information for forecasting air pollution. However, the training process for air quality monitoring models heavily relies on high-performance computing resources, leading to significant energy consumption and associated carbon emissions. This contradicts the objectives of low-carbon and sustainable computing. This paper proposes a New Hybrid PM2.5 Prediction Model (NHPPM) for air quality monitoring to address the above challenges. NHPPM prioritizes energy efficiency while maintaining high prediction accuracy by integrating several power-efficient strategies. Firstly, Wiener filtering is used to denoise multi-source air quality data, enhancing the efficiency of multi-source data fusion. Secondly, Variational Mode Decomposition (VMD) decomposes different components of multi-source air quality data, helping to identify and separate the most important factors affecting pollutants. This reduces the data needed for model training and leads to lower resource consumption. Kernel Principal Component Analysis (KPCA) transforms high-dimensional data into a lower-dimensional representation while retaining critical information, further minimizing computational demands. Additionally, this paper utilizes the Informer deep learning model to analyze trends in air quality data. The model’s effectiveness is validated through ablation studies, performance evaluation experiments, and short-and long-term prediction experiments. The experimental results show that our model reduces the Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) by 16.2% and 14.9%, respectively, compared to existing PM2.5 prediction models. Furthermore, it reduces the energy consumption of model training by 33.8%.