AI-Driven Approaches to Enhancing Energy Efficiencyin IoT Networks for Sustainable Performance

Authors

  • Delshi Howsalya Devi R Author
  • Gunapriya D Author
  • Sivaranjani S Author
  • Suganya N Author
  • Murugesan M Author
  • Anand K Author

Keywords:

Internet of Things, Energy Efficiency, Deep Learning,Reinforcement Learning, Sustainable Computing, Edge Computing, GradientBoosting, Smart Infrastructure

Abstract

Energy efficiency in Internet of Things (IoT) networks represents a criticalchallenge as deployments scale exponentially, with battery-powered devicesrequiring sophisticated optimization strategies to minimise operational costsand environmental impact. This study develops and validates artificial intelligence(AI)-driven methods for energy optimization across heterogeneousIoT environments. We collected 2.4 million temporal data points from 150devices spanning smart homes, industrial facilities, and healthcare systemsover a six-month observation period (January–June 2024). Three algorithmicparadigms—Multi-Layer Perceptron Neural Networks (MLP-NN), GradientBoosting Machines (GBM), and Deep Q-Network Reinforcement Learning(DQN-RL)—were implemented and benchmarked using Mean Absolute Error(MAE), Root Mean Square Error (RMSE), cumulative reward, and energysavingmetrics. The approaches achieved statistically significant energy reductionsof 20–25% across all environments (p <0.001, Cohen’s d = 1.87). GBMattained the lowest prediction error (MAE = 0.03 kWh; RMSE = 0.041 kWh),while DQN-RL demonstrated cumulative rewards of 1487 ± 142and latency reductionsof 35%. These findings establish AI-driven optimization as a scalable,statistically validated solution for sustainable IoT energy management, withimmediate applications in smart infrastructure, Industry 4.0, and precision healthcare.

Downloads

Download data is not yet available.

Downloads

Published

2026-09-22

How to Cite

AI-Driven Approaches to Enhancing Energy Efficiencyin IoT Networks for Sustainable Performance. (2026). Minimax Theory and Its Applications, 11(2s), 461-475. https://journalmta.com/index.php/jmta/article/view/429