Neuro inspired deep learning based secure and energy efficient routing with autonomous intrusion prevention in wireless sensor networks

Abstract

Wireless Sensor Networks (WSNs) are crucial in mission-driven domains such as environmental monitoring, industrial control, and military surveillance. However, their open communication medium, constrained resources, and unattended deployment make them prone to routing-layer attacks. Existing security frameworks mostly rely on reactive intrusion detection systems or conventional deep learning models, which incur high computational overhead and fail to adapt effectively under dynamic network conditions. To overcome these limitations, this study proposes a Neuro-Inspired Deep Learning Framework based on Spiking Neural Networks (SNNs) for autonomous intrusion prevention and energy-aware routing. The proposed model leverages latency-based spike encoding of key behavioral metrics (e.g., residual energy, latency, routing frequency, and packet delivery ratio) and utilizes a Leaky Integrate-and-Fire neuron architecture for proactive vulnerability prediction. Implementation using the Network Simulator-3 (NS-3) simulation tool and validation on the Wireless Sensor Network Dataset (WSN-DS), the framework achieves 99.72 % prediction accuracy, 99.98 % precision, 99.33 % recall, and 99.12 % F1-score, outperforming existing studies in attack detection rate. The proposed Secure Energy-Aware Routing Metric (SEARM) protocol achieves an average energy consumption of 0.32 J and a packet delivery ratio of 99.1 % while maintaining performance across varying network sizes (30–150 nodes) and attack intensities (up to 50 %). Additionally, the model features a self-healing mechanism that reintegrates previously blocked nodes based on dynamic trust recovery. This research establishes a proactive, low-power, and intelligent security paradigm for WSNs and sets the foundation for future innovations in biologically inspired and scalable network protection strategies.