Journal of Positive School Psychology http://journalppw.com
2022, Vol. 6, No. 4, 473-485
© 2021 JPPW. All rights reserved
Design Machine Learning BasedIntelligent Techniques for Detecting Network
Attacks
Shekjavid Hussain
1
; Dr. Bechoo Lal
2
1
Research Scholar, Department of Computer science & Engineering; Shri JJT University Jhunjhunu
Rajasthan
2
Assistant Professor, Department of Computer science & Engineering; Shri JJT University Jhunjhunu
Rajasthan
Abstract
The number of Internet of Things (IoT) devices that are vulnerable to cyber-attacks is
increasing at an alarming rate. As a result, network operators are placing an increasing
emphasis on the control of these devices. A comprehensive packet inspection in
software can be difficult, expensive, rigid, and unable to scale with current network
monitoring solutions that use specialised acceleration on network switches. SDN and
machine learning are used in this work to take use of the programmability offered by
SDNs.
Information driven models for overseeing IoT gadgets in light of their organization
exercises by means of stream based telemetry. The three manners by which we have an
effect: Over a six-month time frame, we gathered traffic follows from 17 genuine
purchaser IoT gadgets and recognized a bunch of traffic streams (per-gadget) that
portray the organization conduct of different IoT gadget types and their working states
(i.e., booting, effectively collaborated with client, or being inactive). (2) We create a
multi-stage design of surmising models that utilization stream levity information to make
forecasts about the organization conduct of different IoT gadget types and their working
states. (3) We measure the compromise among execution and cost of our methodology
and clarify how our checking framework can be used in activity to identify conduct
changes, all utilizing genuine traffic information to prepare our models (firmware
overhaul or digital assaults)..
Keywords: WSN, IoT, Cyber-attack, Security, Machine learning,
1. Introduction
Machine learning (ML) and data mining (DM)
methodologies for cyber security applications
were surveyed in this article. As well as many
applications to cyber intrusion detection
challenges, the ML/DM approaches and
techniques are explained. Paper discusses the
difficulty of ML/DM algorithms in terms of
complexity and recommends which strategies to
utilise depending on what kind of cyber problem
you are trying to address.
When it comes to protecting computers against
assault, unauthorised access or modification or
even destruction, cyber security encompasses a
wide range of technology as well as processes.
There are two main types of cyber security
systems: network and computer (host).
Firewalls, antivirus software, and intrusion
detection systems are all included at the very
least in each of these systems (IDS).
Unauthorized use, copying, change, and
destruction of information systems can be
discovered, determined, and identified with the
aid of IDSs [1]. Internal and external invasions
are two types of security breaches that have
occurred (attacks from within the organization).
Cyber analytics that enable IDSs fall into three
broad categories: misuse-based (also known as
signature-based), anomaly-based, and hybrid. By
looking for patterns in previously detected
assaults, misuse-based techniques can identify
potential threats before they can be launched.
There is no need to worry about a flood of false
alarms because they are excellent at detecting
established assaults. There are rules and
signatures in the database that need to be