A MACHINE LEARNING BASED CLASSIFICATION AND PREDICTION TECHNIQUE FOR DDOS ATTACKS

Abstract

The recent proliferation of Internet of Things (IoT) is paving the way for the emergence of smart cities, where
billions of IoT devices are interconnected to provide novel pervasive services and automate our daily life tasks (e.g.,
smart healthcare, smart home). However, as the number of insecure IoT devices continues to grow at a rapid rate,
the impact of Distributed Denial-of- Service (DDoS) attacks is growing rapidly. With the advent of IoT botnets such
as Mirai, the view towards IoT has changed from enabler of smart cities into a powerful amplifying tool for
cyberattacks. This motivates the development of new techniques to provide flexibility and efficiency of decision
making on the attack collaboration in a software defined networks (SDN) context. The new emerging technologies,
such as SDN and blockchain, give rise to new opportunities for secure, low-cost, flexible and efficient DDoS
attacks collaboration for the IoT environment. In this paper, we propose Co-IoT, a blockchain-based framework for
collaborative DDoS attack mitigation; it uses smart contracts (i.e., Ethereum’s smart contracts) in order to
facilitate the attack collaboration among SDN- based domains and transfer attack information’s in a secure, efficient
and decentralized manner. Co- IoT’s implementation is deployed on the Ethereum official test network Ropsten
[1]. The experimental results confirm that Co-IoT achieves flexibility, efficiency, security, cost effectiveness
making it a promising scheme to mitigate DDoS attacks in large scale