Comparative Analysis of Deep Learning Models for Node Prediction in Low-Power Lossy Networks

Theaccurate estimation of energy consumption within loT networks heavily relies on the precise forecasting of node quantities. This research paper tackles this issue by introducing a deep learning-centered methodology to anticipate node numbers utilizing various dataset parameters like hop count, node degree, count of DIO messages, MAC table entries, and routes. Various deep learning models, such as Long Short-Term Memory (LSTM), Artificial Neural Networks (ANN), Feedforward Neural Networks (FNN), and Convolutional Neural Networks (CNN), are evaluated for their effectiveness in this context. Our findings reveal that ANN outperforms the FNN model, with FNN showing superiority over LSTM in Mean Square Root and R-squared scores. In contrast, CNN exhibits lower performance metrics compared to other models in predicting node quantities, contributing to enhanced energy efficiency and resource distribution in loT setups. This research shifts the focus from conventional security-centric studies towards energy optimization via node prediction, presenting a fresh viewpoint on loT network administration.