Comparison of ML Algorithms to Detect Vulnerabilities of RPL-Based IoT Devices in Intelligent and Fuzzy Systems

Abstract

The RPL protocol (Routing Protocol for Low-Power and Lossy Networks) was designed by IETF [1] for 6LoWPAN to optimize power consumption on the Internet of Things (IoT) devices. These devices have limited processing power, memory, and generally limited energy because they are battery-powered. RPL aims to establish the shortest distance by setting up n number of IoT devices through each other DAG (Directed Acyclic Graph) and therefore the most optimum energy consumption. However, due to the complex infrastructure of RPL and the low capacity of IoT devices, the RPL protocol operating at the network layer is susceptible to attacks. Therefore, it is vital to develop a fast, practical, uncomplicated, and reliable intrusion detection system in the network layer. In the event of an attack on IoT devices operating with the RPL protocol, an anomaly will occur in the network packets in the 3rd layer. Processing these packages with machine learning algorithms will make the detection of the attack extremely easy. In this article, "Decision Tree," (DT) "Logistic Regression," (LR) "Random Forest," (RF) "Fuzzy Pattern Tree Classifier," (FPTC), and "Neural Network" (NN) algorithms are compared for catching Flooding Attacks (FA), Version Number Increase Attacks (VNIA), and Decreased Rank (DRA) attacks. At the end of our study, it is observed that the Random forest algorithm gave better results than other algorithms in the system built by the study.

Distributed and Asynchronous Population-Based Optimization Applied to the Optimal Design of Fuzzy Controllers

“…This fact opens the possibility of applying it to relatively complicated problems to bring down the time required to obtain a solution by (roughly) an order of magnitude or, in a cloud environment, to reduce its cost; this will make this kind of system affordable to small and medium-sized enterprises who will be able to leverage it to add value to their portfolio. Since fuzzy controllers are used extensively on the Internet of Things [52], this could be an excellent area of application. This is left, however, as future work.…”

Prediction of Compromised IoT Infrastructure Using Machine Learning

“…Consequently, the decision tree -based approach efficiently processes and examines IoT data generated by the COOJA simulator to swiftly identify irregular actions and characterize detrimental behavior. According to [11], the Decision Tree, Logistic Regression, Random Forest, Fuzzy Pattern Tree, and Neural Network techniques demonstrated similar outcomes, with Random Forest showing superior overall accuracy. In a different study noted in [15], a machine learning approach centered on Kernel Density Estimation (KDE) was devised to identify Hello Flood ( HF) instances in RPL, achieving an average true positive rate of 84.91% and less than 0.5% false positive rate.…”

“…The dataset was first adjusted to provide an equal amount of normal and impaired flow data. Following that, the dataset was divided into two-thirds test and training datasets (2/3 training, 1/3 testing) [11]. Following this, the data sets were trained and evaluated using several ML algorithms for the detection of DOS attacks (Flooding).…”

“…RPL is specifically engineered to establish DAGs within IoT devices. DAGs consist of a composition of DODAGs (Destination Oriented DAG), with each node in a DODAG striving towards a singular objective [11].…”

“…The root node replies with a DAO-ACK message to all nodes, approving their inclusion. Following this step, nodes with the lowest rank assume the role of the "root," and the aforementioned processes continue [11].…”