Enhancing IoT security: a trust-centric random forest approach for blackhole node detection

The Proposed Model enhances node classification and security in IoT networks by incorporating a Random Forest algorithm to classify nodes as either blackhole or non-blackhole nodes. The Proposed Model utilizes a diverse range of metrics including Packet Delivery Ratio (PDR), Packet Loss, Throughput, Jitter, Latency, Reliability, and Energy Consumption. These metrics are used to train the Random Forest classifier, allowing for accurate identification of malicious nodes attempting blackhole attacks. The Random Forest algorithm, a powerful ensemble learning technique, creates multiple decision trees based on the trust metrics, each making an independent classification. The final classification is determined by majority voting across all the trees. This process improves the detection accuracy by taking into account multiple metrics rather than relying on a single parameter. The combination of this algorithm with the comprehensive trust metrics allows the Proposed Model to better differentiate between benign and malicious nodes, ensuring that only trustworthy nodes are chosen for routing. Simulations were performed to evaluate the performance of the Proposed Model in comparison to RPL and the trust-based approach by Mustafa Ghaleb et al. (2021). The simulations demonstrated that the Random Forest-enhanced model achieved a higher detection accuracy of blackhole nodes, resulting in improved Packet Delivery Ratio (PDR) and throughput. Additionally, the Proposed Model exhibited lower average delay compared to the traditional approaches. By leveraging the Random Forest algorithm, the Proposed Model can more effectively detect and isolate suspicious nodes, preventing them from disrupting network operations. This ensures that the network is both secure and efficient, making it particularly suited for critical IoT applications such as military and battlefield scenarios.