Abstract

System on Chip - Mobile Ad Hoc Networks (SoC-MANETs) are characterized by their decentralized architecture, node mobility, and frequent topology changes. Ensuring secure and reliable service discovery in such environments is critical but challenging due to node misbehavior and trust uncertainties. Traditional trust management mechanisms often fall short in detecting malicious behavior and adapting to network dynamics, especially under unreliable service conditions. This paper proposes a Deep Self-Organizing Control (Deep SOC)-based trust management framework that integrates deep learning with adaptive behavior profiling to enhance the reliability of service discovery. A Convolutional Neural Network (CNN) is employed to predict node trustworthiness based on real-time communication patterns, mobility behavior, and packet integrity. The proposed Deep SOC model was tested using NS-3 simulation with 100 nodes under varying mobility and attack scenarios. It achieved a Packet Delivery Ratio (PDR) of 92.8%, End-to-End Delay (E2ED) of 116 ms, Detection Accuracy of 95.4%, Trust Convergence Time of 8.4s, and Energy Consumption of 21.3J outperforming existing methods by 12–18% across metrics.

Authors

P. Ramya¹, S. Venkatesh Babu², D. Jebakumar Immanuel³
PSNA College of Engineering and Technology, India¹, KGISL Institute of Technology, India², Karpagam Institute of Technology, India³

Keywords

SoC-MANET, Trust Management, Deep Learning, Service Discovery, Network Security