Hybrid Temporal-Spatial Energy Prediction in Wireless Sensor Networks using Transformer and Graph Neural Network Models

Wireless Sensor Networks (WSNs) are widely used in a variety of applications, but energy efficiency is still a major concern. Accurate energy prediction can dramatically improve network lifetime and performance. This study investigates energy prediction in WSNs with transformer-based models and graph neural networks (GNNs). Our tests show that our hybrid model improves network lifespan by 15%, with a Mean Squared Error (MSE) of 0.034 and an R-squared score of 0.92. It does this better than baseline models like LSTM and standalone GNNs. The suggested method takes into account both time and space dependencies, which leads to better resource allocation and a longer network lifespan.