Registrujte se

Bezpečnost a soukromí

Soukromí, šifrování, sledování, záloha dat, monitoring, audit, hesla, hacking, cracking, malware, phishing

Saving the World: Increasing Efficiency and Accuracy of Encrypted Traffic Analysis of People at Risk

Přednáška | D105 | Neděle 11:00 - 11:45 |

Activists, journalists and human rights defenders are in hostile environments and in constant danger as they deal with sensitive information. They are often exposed to targeted and sophisticated attacks. We designed the Emergency VPN which allows us to help people in danger by analyzing their mobile traffic. This way we can identify if a device is infected or find its vulnerabilities that may put the user at risk. However, the biggest challenge for the network analyst is to quickly and accurately detect malicious encrypted traffic. The speed of the analysis is a critical factor in this work. To improve the speed of the analysis of HTTPS traffic, we combine specific features extracted from HTTPS traffic with state of the art machine learning methods. In this talk we will show how this combination allowed us to increase the efficiency and accuracy of Encrypted traffic analysis of people at risk. In a live demo, we will demonstrate a detection of malicious traffic in a mobile device.

Bezpečnost a soukromí Security Data Bezpečnost a soukromí HTTPS TLS Network Security Malware TLS 1.2 TLS 1.3 Machine Learning Traffic Civil society Hacktivity

František Střasák

Centrum Umělé Inteligence FEL ČVUT

Jan Fajfer

Centrum Umělé Inteligence ČVUT

Veronica Valeros

Centrum Umělé Inteligence ČVUT

Fantastic Attacks and How Kalipso can Find Them

Přednáška | D105 | Neděle 13:00 - 13:45 |

Detecting attacks in a network is very hard due to the huge amount of information, and the similarity between attacks and normal traffic. Knowing the traffic of your computer is hard enough, more so in a large network. An analyst has to decide and block infected computers without being aware of all the details. A company may afford a large detection system based on big data, but what about you? Slips is a network intrusion detection system that uses flows, behaviors, and machine learning to detect attacks in a network. Based on Zeek and with a modular structure it is easy to extend the system with new models of your design, leaving the final decision to an internal ensembling algorithm. From flow-based port scan detection to anomaly detection, threat intelligence, VirusTotal integration, geolocation and machine learning profiling, slips includes modules that can give a comprehensive high-level view of your security. However, it is very hard to show this information clearly and to include the analyst in the process. Enter Kalipso. Kalipso is a nodejs-based terminal interface designed to display the complexity of the information produced by Slips. This interface helps traffic analysts to quickly get a superficial understanding of what is going on in the network. With animated graphs and charts based on the blessed and the blessed-contrib libraries, it is possible to configure and connect data from Slips meaningfully. After slips filled the redis database, Kalipso is ready to display the information. It creates a tree with all IP addresses in the traffic, separating the data in time windows. For every IP and time window, it shows a timeline, detections, and a map with the geolocation of all the destination IPs contacted. Each IP address is modeled using stacked bars and tables based on the destination ports contacted, destination IPs contacted, source ports used, and ports opened as a server. Different windows are accessed with hotkeys, and important information is highlighted with several font types and colors. Distinctive outgoing connections are displayed together with their VirusTotal information and behavioral model. Complete with the ability to copy information to the clipboard or save it into a file, Kalipso allows the analyst to rapidly overview what is happening in a network.

Bezpečnost a soukromí Security Bezpečnost a soukromí Attacks machine learning nodejs Interface Python Free software malware OWASP

Sebastian Garcia


Kamila Babayeva