Machine Learning Algorithms for Preventing IoT Cybersecurity Attacks
The goal of this paper is to understand the effectiveness of machine learning (ML) algorithms in combatting IoT-related cyber-attacks, with a focus on Denial of Service (DoS) attacks. This paper also explores the overall vulnerabilities of IoT devices to
- PDF / 352,654 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 121 Downloads / 202 Views
Abstract. The goal of this paper is to understand the effectiveness of machine learning (ML) algorithms in combatting IoT-related cyber-attacks, with a focus on Denial of Service (DoS) attacks. This paper also explores the overall vulnerabilities of IoT devices to cyber-attacks, and it investigates other datasets that can be used for IoT cyber-defense analysis, using ML techniques. Finally, this paper presents an evaluation of the CICDoS2019 dataset, using the Logistic Regression (LR) algorithm. With this algorithm, a prediction accuracy of 0.997 was achieved. Keywords: Cybersecurity · Internet of Things · Machine learning (ML) · Supervised Learning · Unsupervised Learning · Reinforcement Learning (RL) · Logistic Regression (LR) · DDoS · Botnet
1 Purpose In recent years there has been a proliferation of devices that use the Internet for communications. These devices have been given the name IoT (Internet of Things) because they extend the traditional form-factor of computing devices to forms that are more amenable to diverse use-cases. IoT devices encompass a wide range of devices, such as sensors, actuators, smart-devices, RFID-devices, etc. [1] and they are all connected to the Internet for the purposes of collecting and transmitting data. The number of IoT devices connected to the Internet has grown to the point that by 2015 there were over 4.9 billion devices connected. By 2020, there will be 25 billion connected IoT devices. This growth has brought about an enormous amount data and network traffic. This data can be classified as Big Data because of the volume of data that has been generated (terabytes and greater), the variety of data (NetFlow, pcap, etc.) and because of the velocity (rate of change of collection) of the data. Although these devices have brought about major advances, they have also brought about new challenges: they are the new frontier for cyber-attacks. IoT devices are vulnerable to cyber-attacks because IoT devices are heterogeneous (different types, different methods of communication and different types of data being transmitted), they are numerous (in the billions), they have limited computing resources and because they normally operate on the edge of computer networks. © Springer Nature Switzerland AG 2021 K. Arai et al. (Eds.): IntelliSys 2020, AISC 1252, pp. 679–686, 2021. https://doi.org/10.1007/978-3-030-55190-2_53
680
S. Chesney et al.
Distributed Denial of Service attacks (DDoS) are one example of the type of cyberattacks that IoT devices will need to be protected against. Since traditional cyber security methods are ineffective for IoT device, ML-enabled solutions are the new means of protecting IoT devices. With the many ML algorithms that are available, the Logistic Regression (LR) algorithm was used in this project against the CICDDoS2019 [2] dataset. This dataset has been made publicly available for research and testing by the University of New Brunswick and the Canadian Institute of Cybersecurity (CIC). In light of this information, this paper has been organized as follo
Data Loading...
