Measurement: Sensors (Jun 2024)
Ensemble classification to predict botnet and its impact on IoT networks
Abstract
The IoT (Internet of Things) has been rapidly growing to make a stronger influence on huge industrial systems. Since cybercriminals have made IoT a target for harmful operations such as botnets, an attack on the end nodes is now a possibility. Protecting IoT infrastructure with a typical intrusion detection system is difficult because of its vastness, variety, and minimal resource availability. Harmful ''bot sources” (servers) control a botnet, which is made up of hacked devices that are employed for several illicit purposes like sending spam, initiating DoS attacks, and stealing personal data. Because bot sources generate network traffic while conversing with their bots, Intrusion Detection Systems could benefit from analyzing traffic on the network to detect Botnet traffic. DDoS attacks and spam distribution are common uses for botnets.As machine learning (ML) approaches are employed in various niches of security, it appears practical and workable to use ML to detect botnets. ML has been used in various researches to detect botnets. However, the results are either unreliable or limited to certain types of botnets or devices. Current approaches attempt to solve these issues by presenting models that are trained on botnet features, but the large dimensionality of feature values and the reliance on botnet features alone are serious drawbacks. In order to address these constraints, a novel botnet and its impact detection by a typical ensemble classification approach has been proposed by the current work. The framework uses the correlated traffic-flow and botnet features to train the classifier, which is titled as ''Ensemble Classification to Predict Botnet and its Impact (EC-PBI) on IoT Networks''. The experimental study is a cross validation approach that is performed using a multi-label, fourfold strategy. Performance analysis of the proposed approach was done by comparing it with contemporary models. Results prove its efficiency in detecting botnet and its impact.
