Abstract. Most networks are connected to the Internet through firewalls to block attacks from the outside and to limit communication initiated from the inside. Because of the limited, supposedly safe functionality of the Domain Name System protocol, its traffic is by and large neglected by firewalls. The resulting possibility for setting up information channels through DNS tunnels is already known, but all existing implementations require help from insiders to set up the tunnels. This paper presents a new Metasploit module for integrated penetration testing of DNS tunnels and uses that module to evaluate the potential of DNS tunnels as communication channels set up through standard, existing exploits and supporting many different command-and-control malware modules.

No abstract

Malware detection techniques come in many different flavors, and cover different effectiveness and efficiency trade-offs. This paper evaluates a number of machine learning techniques to combine multiple static Android malware detection techniques using automatically constructed decision trees. We identify the best methods to construct the trees. We demonstrate that those trees classify sample apps better and faster than individual techniques alone.

No abstract

No abstract