Next generation AI powered framework for autonomous energy optimization and real time anomaly detection in IoT driven...

Area of Science:

• Computer Science
• Network Engineering
• Artificial Intelligence

Background:

• Internet of Things (IoT)-based Wireless Sensor Networks (WSNs) are crucial across sectors, necessitating robust solutions for energy efficiency and security.
• Traditional WSN methods struggle with real-time data processing, dynamic network adaptation, and evolving anomalies, impacting performance and dependability.
• Existing approaches fall short in addressing the complex challenges of energy consumption and anomaly detection in large-scale, dynamic WSN environments.

Purpose of the Study:

• To introduce LEGO-WSN, a novel intelligent solution designed to enhance energy efficiency and enable real-time faulty node identification in WSNs.
• To leverage the combined power of Long Short-Term Memory (LSTM) with an attention mechanism and Genetic Algorithm (GA) optimization for improved WSN performance.
• To address the limitations of traditional methods by providing a scalable and adaptable solution for WSN energy optimization and anomaly detection.

Main Methods:

• Proposed LEGO-WSN integrates LSTM with an attention layer and GA for fault diagnosis and anomaly detection in WSNs.
• The Genetic Algorithm (GA) is employed to optimize network transmission parameters and sensor operational planning.
• LSTM, enhanced with attention mechanisms, analyzes time-series data to identify patterns indicative of network anomalies, such as blackhole attacks.

Main Results:

• LEGO-WSN achieved a 20% reduction in energy consumption, demonstrating significant energy efficiency improvements.
• The system exhibited high accuracy in real-time anomaly detection, reaching 99% accuracy, 98% precision, and 99% recall.
• Evaluated on real-world WSN data, the model showed flexibility across diverse environmental and network conditions.

Conclusions:

• LEGO-WSN presents a novel, scalable, and reliable solution for optimizing WSN performance.
• The proposed approach effectively enhances both the energy efficiency and security of WSNs.
• This intelligent system addresses key challenges in WSNs, offering substantial improvements over traditional methods.