Single Attack to Multi-attack Intrusion Detection System- A Survey

  • S. R. Khonde, V. Ulagamuthalvi


In new era due to huge use of internet most of the networks are getting compromised. Main reason behind it is an intelligent system which generates some mischief data which easily breaks security of network and become vulnerable. Such systems or a mankind is called as intruder.  Most of the intruders break security of network and compromise network such that it will enter in an unstable or unsecure state. Intruders make system vulnerable such that easily all activities from network can be controlled and managed. Any vulnerable system is open for any type of malicious activity as no security breach is available. These types of systems mostly targeted by attack a type of malicious activity. Attacks once happened on the network or a system mostly captures all the important information about the network to conduct deceitful activities. Once the intrusion happen in system it is difficult to stop and rectify it. To avoid such type of attacks an intelligent device is used called intrusion detection system (IDS). IDS mostly helps administrator to avoid malicious activities to happen or enter in the network. Most of the IDS nowadays are using various machine learning techniques to detect and stop such type of malicious activities. Most of the IDS use to detect single type of attack using various algorithms. Now a day’s intruders are trying to enter into the network by changing the behaviour of attack from single to multi. At a time if system is compromised with multi attack then it is not possible for any IDS to detect it and stop it. Multi-attack can be defined as a multiple attacks attacking on the system at same time or combination of various attacks attacking on system. To handle such type of attack IDS systems need to improve so that they can detect multi-attack with single attack. Standard dataset used by IDS also provides signature of single attack. None of the standard dataset provides or handles signatures of multi-attack. In real work most of the attack happening now a days are not basic attacks neither single attack. To provide better security need to IDS is to detect multi-attack. In this paper, we discuss various types of attacks and approaches to implement intelligent IDS. This paper also focuses on various machine learning algorithms which can be used for attack detection. Paper also provides survey of various datasets used for attack detection by IDS.