Fog-Based Deep Learning for Real-Time Cold Chain Temperature Prediction Using IoT Data

A third of the food produced globally and in South Africa is lost or wasted annually. Fresh fruits and vegetables (FFVs) contribute 44% of South Africa’s wastage, with temperature abuse as the main cause due to their high sensitivity and perishability. Real-time cold chain man-agement with predictive analytics is necessary to control parameters and minimise temperature breaks proactively. While Machine Learning (ML) can predict temperature, cloud deployment causes latency, bandwidth demands, and internet dependency, hindering real-time operations. Fog computing mitigates this by localising ML predictions, an area underexplored for the FFV cold chain. This study investigates fog-based Deep Neural Networks for predicting cold room temperatures in FFV cold chains, utilising IoT data from a South African apple cold room laboratory. SimPy, MinMax scaling, 75%/25% train/test splitting and a sequence length of 2 were used. The fog-based LSTM and GRU ensemble outperformed the cloud-based model with R2 [0.9081, 0.9245] vs [0.5477, 0.5992], MAE [0.9999 o C, 1.6353 o C] vs [4.2983 o C, 4.6029 o C], MSE [10.0084,12.0685] vs [50.6492, 56.4827], processing time [0.0840, 0.0847] vs [0.3375, 0.3378], and Pearson R [0.9533, 0.9618] vs [0.7670, 0.7930] (95% confidence). Paired t-test and Wilcoxon test confirm fog’s significant superiority. Both had high data utilisation of 99.8%, guaranteeing analysis for transmitted data. High accuracy and low processing time make fog ideal for real-time cold chains, reducing FFV wastage while improving sustainability, affordability, profitability, and food security. Future work will add sensors, fog nodes, compare models, consider different datasets and implement asynchronous sensor fusion and temperature break and cause predictions.