Extended Game Theoretic Dirichlet Based Collaborative Intrusion Detection Systems
Security has always been one of the key issues of any man-made system, this paved the way for a submodule or application or a device to monitor or system for malicious activities. This system or submodule or device is known as Intrusion Detection System (
- PDF / 274,561 Bytes
- 14 Pages / 439.37 x 666.142 pts Page_size
- 36 Downloads / 191 Views
Abstract Security has always been one of the key issues of any man-made system, this paved the way for a submodule or application or a device to monitor or system for malicious activities. This system or submodule or device is known as Intrusion Detection System (IDS). As technology evolves so does the associated threats and thus the intrusion detection system needs to evolve. Game theory throws in a different perspective which have not been looked upon much. Game theory provides a way of mathematically formalizing the decision making process of policy establishment and execution. Notion of game theory can be used in intrusion detection system in assisting in defining and reconfiguring security policies given the severity of attacks dynamically. We are trying to formulate a robust model for the theoretical limits of a game theoretic approach to IDS. The most important flaw of game theory is that it assumes the adversary’s rationality and doesn’t take into consideration multiple simultaneous attacks. Therefore, a collaborative trust and Dirichlet distribution based robust game theoretic approach is proposed which will try to resolve this issue. Reinforced learning approaches using Markov Decision Process will be utilized to make it robust to multiple simultaneous attacks.
Keywords Intrusion detection system Dirichlet based trust management Collaborative trust management Game theory Nash equilibrium
S. Paul (&) T. Makkar K. Chandrasekaran Department of Computer Science and Engineering, National Institute of Technology Karnataka, Surathkal, India e-mail: [email protected] T. Makkar e-mail: [email protected] K. Chandrasekaran e-mail: [email protected] © Springer Science+Business Media Singapore 2016 M. Senthilkumar et al. (eds.), Computational Intelligence, Cyber Security and Computational Models, Advances in Intelligent Systems and Computing 412, DOI 10.1007/978-981-10-0251-9_32
335
336
S. Paul et al.
1 Introduction Intrusion Detection Systems plays a key role in security of modern day software applications/systems. They compare observable behaviour in the system against suspicious patterns to identify any kind of intrusions. There are two variances of IDS: Network based (NIDS) or Host based (HIDS). Traditional IDSs have a problem that they work in isolation and therefore have higher chance of getting compromised by unknown or new threats. A Collaborative IDS solves this problem by having peer IDS help each other out and get aided by shared collective knowledge and experience from peers. This increases both the accuracy and the ability to detect new intrusion threats. Collaborative IDS assumes that all IDSs will honestly cooperate. The lack of trust management leaves the system vulnerable to malicious peers [1]. Few IDSs have been produced to cooperate honestly based on trust and/or distributed trust models but they have not incorporated any kind of incentives for IDS collaboration. Incentives are important criteria any collaborative system otherwise it will suffer from “free rider probl
Data Loading...
