Passive Visual Fingerprinting of Network Attack Tools Gregory Conti Kulsoom Abdullah College of Computing Georgia Institute of Technology Passive Visual Fingerprinting of Network Attack Tools Gregory Conti Kulsoom Abdullah College of Computing Georgia Institute of Technology Nessus 2.0.10, IP to Port to Port to IP
Motivation Common network reconnaissance and vulnerability assessment tools can be visualized in such a way as to identify the attack tool used. Law enforcement forensics Identify characteristics of new tools/worms Provide insight into attacker’s methodology & experience level Help network defender to initiate appropriate response
System Architecture Ethernet Packet Capture Parse Process Plot tcpdump (pcap, snort) Perl xmgrace (gnuplot) winpcap VS tcpdump capture files Packet Capture Parse Process Plot Interact
Examining Available Data… Link Layer (Ethernet) All raw data available on the wire: Application layer data Transport layer header Network layer header Link layer header Network Layer (IP) Focused on: Source / Destination Port Source / Destination IP Timestamp Length of raw packet Protocol Type Transport Layer (TCP) IP: http://www.ietf.org/rfc/rfc0791.txt UDP: http://www.ietf.org/rfc/rfc0768.txt TCP: http://www.ietf.org/rfc/rfc793.txt Transport Layer (UDP) Ethernet: http://www.itec.suny.edu/scsys/vms/OVMSDOC073/V73/6136/ZK-3743A.gif
Attacks Fingerprinted nessus 2.0.10 nmap 3.0 nmap 3.5 nmapwin 1.3.1 Superscan 3.0 Superscan 4.0 nikto 1.32 scanline 1.01 sara 5.0.3 NSA CDX dataset 2003 http://www.insecure.org/tools.html
Visualizations Time Sequence Data Port and IP Mapping Sequence of Source/Destination Ports and IP’s Sequence of Packet Lengths Sequence of Packet Protocols Port and IP Mapping Source Port to Destination Port Source IP to Destination IP Source IP to Destination Port Source Port/IP to Destination IP/Port Source IP/Port to Destination Port/IP Characterization of home/external network Fixed memory requirements
parallel plot views External IP Internal IP 255.255.255.255 255.255.255.255 0.0.0.0 0.0.0.0 External Port Internal Port 65,535 65,535 0 0 External IP Internal Port 255.255.255.255 65,535 0.0.0.0 0
Baseline External Port Internal Port External IP Internal IP
nmap 3 (RH8) nmap 3 UDP (RH8) scanline 1.01 (XP) SuperScan 3.0 (XP) Using (mostly) default scan options TCP = Green UDP = Orange Foundstone… superscan family, scanline NMapWin… runs on an nmap 3 engine NMapWin 3 (XP) nmap 3.5 (XP) nikto 1.32 (XP) SuperScan 4.0 (XP)
Sara 5.0.3 (port to port) Light Medium Heavy
Georgia Tech Honeynet External IP Internal Port External Port Internal Port External IP Internal IP
Also a Port to IP to IP to Port View External IP External Port Internal Port Internal IP 255.255.255.255 65,535 65,535 255.255.255.255 0.0.0.0 0 0 0.0.0.0 Also a Port to IP to IP to Port View
Exploring nmap 3.0 in depth (port to IP to IP to port) default (root) stealth FIN (-sF) NULL (-sN) UDP (-sU) SYN (-sS -O) stealth SYN (-sS) CONNECT (-sT) XMAS (-sX)
nmap within Nessus (port to IP to IP to port) CONNECT (-sT) Nessus 2.0.10 UDP (-sU)
SuperScan Evolution (port to IP to IP to port) scanline 1.01
packet length and protocol type over time packets ports length
WinNMap Compress the time domain to distill sequence
SuperScan 4.0
time sequence data (external port vs. packet) nmap win superscan 3 ports ports packets packets Also internal/external IP and internal port
tool interface
Findings (Weaknesses) Interaction with personal firewalls Countermeasures Scale / labeling are issues Occlusion is a problem Greater interactivity required for forensics and less aggressive attacks Some tools are very flexible Source code not available for some tools
Findings (Strengths) Aggressive tools have distinct visual signatures Threading / multiple processes may be visible Some source code lineage may be visible Some OS/Application features are visible Some classes of stealthy attack are visible
Findings (Strengths) Sequence of ports scanned visible Frequently attacked ports visible Resistant to high volume network traffic Viable in the presence of routine traffic Useful against slow scans (hours-weeks) Useful against distributed scans
Future Work Add forensic capability Task driven interactivity (Zoom & filter, details on demand) Smart books (images & movies) Usability studies Stress test Explore less aggressive attack classes
Demo
classic infovis survey security infovis survey rumint tool http://www.rumint.com/software.html classic infovis survey www.cc.gatech.edu/~conti security infovis survey www.cc.gatech.edu/~conti VizSEC Paper/Slides http://users.ece.gatech.edu/~kulsoom/research.html www.cc.gatech.edu/~conti Visual Security Community http://www.ninjabi.net/index.php?option=com_nxtlinks& catid=41&Itemid=47 Kulsoom’s Research http://users.ece.gatech.edu/~kulsoom/research.html
Acknowledgements Dr. John Stasko Dr. Wenke Lee Dr. John Levine http://www.cc.gatech.edu/~john.stasko/ Dr. Wenke Lee http://www.cc.gatech.edu/~wenke/ Dr. John Levine http://www.eecs.usma.edu/ Julian Grizzard http://www.ece.gatech.edu/ 404.se2600 Clint Hendrick icer Rockit StricK
Questions? Greg Conti conti@cc.gatech.edu www.cc.gatech.edu/~conti Kulsoom Abdullah gte369k@mail.gatech.edu http://users.ece.gatech.edu/~kulsoom/research.html Image: http://altura.speedera.net/ccimg.catalogcity.com/210000/211700/211780/Products/6203927.jpg