Biblio

Cyber risk assessments are an essential process for analyzing and prioritizing security issues. Unfortunately, many risk assessment methodologies are marred by human subjectivity, resulting in non-repeatable, inconsistent findings. The absence of repeatable and consistent results can lead to suboptimal decision making with respect to cyber risk reduction. There is a pressing need to reduce cyber risk assessment uncertainty by using tools that use well defined inputs, producing well defined results. This paper presents Automated Vulnerability and Risk Analysis (AVRA), an end-to-end process and tool for identifying and exploiting vulnerabilities, designed for use in cyber risk assessments. The approach presented is more comprehensive than traditional vulnerability scans due to its analysis of an entire network, integrating both host and network information. AVRA automatically generates a detailed model of the network and its individual components, which is used to create an attack graph. Then, AVRA follows individual attack paths, automatically launching exploits to reach a particular objective. AVRA was successfully tested within a virtual environment to demonstrate practicality and usability. The presented approach and resulting system enhances the cyber risk assessment process through rigor, repeatability, and objectivity.

The number of vulnerabilities and reported attacks on Web systems are showing increasing trends, which clearly illustrate the need for better understanding of malicious cyber activities. In this paper we use clustering to classify attacker activities aimed at Web systems. The empirical analysis is based on four datasets, each in duration of several months, collected by high-interaction honey pots. The results show that behavioral clustering analysis can be used to distinguish between attack sessions and vulnerability scan sessions. However, the performance heavily depends on the dataset. Furthermore, the results show that attacks differ from vulnerability scans in a small number of features (i.e., session characteristics). Specifically, for each dataset, the best feature selection method (in terms of the high probability of detection and low probability of false alarm) selects only three features and results into three to four clusters, significantly improving the performance of clustering compared to the case when all features are used. The best subset of features and the extent of the improvement, however, also depend on the dataset.