A SECURE FEDERATED INTRUSION DETECTION MODEL WITH BLOCKCHAIN AND DEEP BIDIRECTIONAL (LSTM)

Abstract

Several specialists have suggested using Machine Learning (ML) methods to create Intrusion
Detection Systems (IDMs) in order to enhance the security of computing and network resources. Machine
Learning replicates human cognitive skills by modeling deduction, inference, extrapolation, and synthesis. The
use of this technology might be employed to construct Intrusion Detection Systems (IDMs), thereby facilitating
accurate detection of harmful network traffic, resulting in a reduction in false positive alarms. Support Vector
Machines (SVM), Decision Trees (DT), Bayesian Networks, and Naïve Bayes are often used machine learning
techniques for creating Intrusion Detection Systems (IDMs). Nevertheless, the aforementioned conventional
machine learning (ML) techniques prioritize the labor-intensive process of designing features and are less
effective when confronted with vast amounts of data that need categorization in a real-world setting. As the
amount of the dataset increases, the accuracy of many categorization jobs decreases. To address the difficulties
in analyzing large amounts of data that classical machine learning (ML) algorithms face, deep learning (DL)
methods are used to enhance the efficiency of intelligent data mining (IDM), particularly in situations involving
many classes.
An intrusion refers to any illegal action that results in damage to a computer system or network, with the
potential to jeopardize the confidentiality, integrity, or availability of information. Intrusion detection models
are often integrated into software programs to monitor networks or computers for malicious activity in order to
maintain system security [10]. As a proactive solution, IDM detects and halts possible risks to a system or
network before they may do harm, including both hostile insiders and external hackers.