Title: Botnet attack detection in IoT using hybrid optimisation enabled deep stacked autoencoder network

Authors: Archana Kalidindi; Mahesh Babu Arrama

Addresses: Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Hyderabad, Telangana, India ' Department of Computer Science and Engineering, Koneru Lakshmaiah Education Foundation, Hyderabad, Telangana, India

Abstract: IoT devices have become a major target for numerous cyber-attacks, among which botnet-based attacks are the most serious and commonly occurring type. Thus, accurate detection schemes have become essential in identifying malicious attacks to safeguard IoT devices against various security threats. This work devises a deep learning (DL)-based botnet attack detection scheme in IoT. Attack detection is accomplished using the help of a deep stack auto encoder (DSA) based on the various features obtained from the traffic data. The salient features in the input data are determined by considering information gain. A novel Adam average subtraction based optimisation (AASBO) algorithm is presented for adjusting the learning parameters and the weights of the DSA. The ASSBO-DSA is evaluated for its effectiveness by considering various measures, such as precision, recall, and F-measure, and it attained a maximal value of 0.884, 0.874, and 0.879, respectively.

Keywords: IoT; botnet attack; information gain; deep stacked auto encoder; DSA; optimisation; Adam average subtraction based optimisation; AASBO.

DOI: 10.1504/IJBIC.2023.134981

International Journal of Bio-Inspired Computation, 2023 Vol.22 No.2, pp.77 - 88

Received: 05 Nov 2022
Accepted: 21 Apr 2023
Published online: 22 Nov 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article