Search for: All records

Internet-wide scanning is a critical tool for security researchers and practitioners alike. By exhaustively exploring the entire IPv4 address space, Internet scanning has driven the development of new security protocols, found and tracked vulnerabilities, improved DDoS defenses, and illuminated global censorship. Unfortunately, the vast scale of the IPv6 address space—340 trillion trillion trillion addresses—precludes exhaustive scanning, necessitating entirely new IPv6-specific scanning methods. As IPv6 adoption continues to grow, developing IPv6 scanning methods is vital for maintaining our capability to comprehensively investigate Internet security. We present 6SENSE, an end-to-end Internet-wide IPv6 scanning system. 6SENSE utilizes reinforcement learning coupled with an online scanner to iteratively reduce the space of possible IPv6 addresses into a tractable scannable subspace, thus discovering new IPv6 Internet hosts. 6SENSE is driven by a set of metrics we identify and define as key for evaluating the generality, diversity, and correctness of IPv6 scanning. We evaluate 6SENSE and prior generative IPv6 discovery methods across these metrics, showing that 6SENSE is able to identify tens of millions of IPv6 hosts, which compared to prior approaches, is up to 3.6x more hosts and 4x more end-site assignments, across a more diverse set of networks. From our analysis, we identify limitations in prior generative approaches that preclude their use for Internet-scale security scans. We also conduct the first Internet-wide scanning-driven security analysis of IPv6 hosts, focusing on TLS certificates unique to IPv6, surveying open ports and security-sensitive services, and identifying potential CVEs. 

Background and Context: GitHub has been recently used in Software Engineering (SE) classes to facilitate collaboration in student team projects as well as help teachers to evaluate the contributions of their students more objectively. Objective: We explore the benefits and drawbacks of using GitHub as a means for team collaboration and performance evaluation in large SE classes. Method: Our research method takes the form of a case study conducted in a senior level SE class with 91 students. Our study also includes entry and exit surveys, an exit interview, and a qualitative analysis of students’ commit behavior. Findings: Different teams adapt GitHub to their workflow differently. Furthermore, despite the steep learning curve, using GitHub should not affect the quality of students’ submissions. However, using GitHub metrics as a proxy for evaluating team performance can be risky. Implications: We provide several recommendations for integrating Web-based configuration management tools in SE classes. 

Whether supernovae are a significant source of dust has been a long-standing debate. The large quantities of dust observed in high-redshift galaxies raise a fundamental question as to the origin of dust in the Universe since stars cannot have evolved to the AGB dust-producing phase in high-redshift galaxies. In contrast, supernovae occur within several millions of years after the onset of star formation. This white paper focuses on dust formation in supernova ejecta with US-Extremely Large Telescope (ELT) perspective during the era of JWST and LSST.