Machine learning for QoS and security enhancement of RPL in IoT-Enabled wireless sensors

Abstract

Internet of Things (IoT) networks rely on wireless sensors for data collection and transmission, making them vulnerable to security threats that undermine their Quality of Service (QoS). The Routing Protocol for Low-Power and Lossy Networks (RPL) is crucial for efficient data transmission in IoT networks, but its performance can be significantly degraded by attacks such as Rank, Sinkhole and Wormhole attacks. These threats disrupt network integrity by manipulating routing information, attracting traffic through malicious nodes and tunneling data to malicious endpoints. This paper presents a novel machine learning-based framework to enhance RPL's security and QoS. Our approach integrates a random forest model for precise traffic classification, a reinforcement learning module for dynamic and adaptive routing, and a modified RPL objective function that incorporates classification outcomes into routing decisions. Simulations demonstrate that our framework significantly improves network throughput, reduces latency, and enhances packet delivery ratios while maintaining low jitter. Furthermore, it achieves a high detection rate, minimal false positives, and swift response to security incidents, thereby robustly securing the RPL protocol and enhancing QoS in IoT-enabled wireless sensor networks. The findings of this research offer substantial contributions to the field, providing a comprehensive solution to strengthen RPL against prevalent security threats.