Energy Conservation for Internet of Things Tracking Applications Using Deep Reinforcement Learning

Area of Science:

• Computer Science
• Artificial Intelligence
• Internet of Things

Background:

• Internet of Things (IoT) systems require efficient sensor networks for applications like smart farms, factories, and cities.
• Battery-powered sensor devices in IoT tracking systems face significant energy consumption challenges.
• Traditional sensor selection methods struggle with the dynamic demands and limited battery life of IoT applications.

Purpose of the Study:

• To develop an energy-efficient sensor selection model for IoT-based target tracking applications.
• To address the limitations of existing methods in managing sensor node battery lifetime.
• To leverage deep reinforcement learning for optimizing sensor selection in real-time tracking.

Main Methods:

• Proposed a novel model combining Long Short-Term Memory (LSTM) with Deep Q-Network (DQN), a type of deep reinforcement learning (Deep RL).
• Implemented a sensor selection strategy based on minimizing a distance function to identify the most energy-efficient sensor.
• Utilized simulation to evaluate the performance of the proposed Deep RL target tracking model.

Main Results:

• The proposed LSTM-DQN model effectively selects the best sensor for target tracking, prioritizing energy efficiency.
• Demonstrated significant improvements in reducing sensor device energy consumption compared to traditional approaches.
• Simulation results indicate favorable outcomes in both sensor selection accuracy and overall energy savings.

Conclusions:

• The LSTM-DQN based Deep RL model offers a promising solution for energy-efficient sensor selection in IoT target tracking.
• This approach effectively mitigates the battery lifetime constraints of sensor nodes in real-time IoT applications.
• The method contributes to the development of more sustainable and long-lasting smart IoT systems.