Nature Inspired Online Real Risk Assessment Models for Security Systems
EuroISI '08 Proceedings of the 1st European Conference on Intelligence and Security Informatics
Combining incremental Hidden Markov Model and Adaboost algorithm for anomaly intrusion detection
Proceedings of the ACM SIGKDD Workshop on CyberSecurity and Intelligence Informatics
| Hi-index | 0.00 |
A Distributed Intrusion Prediction and Prevention Systems (DIPPS) not only detects and prevents possible intrusions but also possesses the capability to predict possible intrusions in a distributed network. Based on the DIPS sensors, instead of merely preventing the attackers or blocking traffic, we propose a fuzzy logic based online risk assessment scheme. The key idea of DIPPS is to protect the network(s) linked to assets, which are considered to be very risky. To implement DIPPS we used a Distributed Intrusion Detection System (DIDS) with extended real time traffic surveillance and online risk assessment. To model and predict the next step of an attacker, we used a Hidden Markov Model (HMM) that captures the interaction between the attacker and the network. The interaction between various DIDS and integration of their output are achieved through a HMM. The novelty of this paper is the detailed development of Fuzzy Logic Controllers to estimate the various risk(s) that are dependent on several other variables based on the inputs from HMM modules and the DIDS agents. To develop the fuzzy risk expert system, if-then fuzzy rules were formulated based on interviews with security experts and network administrators. Preliminary results indicate that such a system is very practical for protecting assets which are prone to attacks or misuse, i.e. highly at risk.