icfs instituto de ciencias forenses y de la seguridad alvaro ortigosa
TRANSCRIPT
ICFSInstituto de Ciencias Forenses y de la Seguridad
Alvaro Ortigosa
Transatlantic Network Security Initiative
Combining Technical Analysis with Social Network Analysis for
an Early Warning System for Cyber Attacks
IP: Sanjay Goel
3
• SCADA systems integrated
• Recent reports claim US power grid compromised
• Possible link between blackouts and terrorists
• Smart Grid cause for concern
SCADA SystemsInfrastructure Risk
4
• There has been a relative lack of forthcoming information on the attacks on the critical infrastructure
• Probing and attacks continue from several sources (nations and transnational groups)
• Attacks on the infrastructure are inevitable
• We need to intelligently manage our risks
SCADA SystemsInfrastructure Risk
5
SCADA SystemsVulnerabilities in Infrastructure
4:10 pm Transmission lines start tripping in Michigan and Ohio blocking flow of power East. Due to deficit Generators shut down. Causing blackout in the East
1:58 pm Eastlake Ohio plant shuts down
3:06 pm A First Energy 345-KV transmission line fails south of Cleveland Ohio
3:17 pm Voltage dips temporarily on Ohio portion of grid causing power to shift to another transmission line which fails
3:41& 3:46 pm 2 breakers connecting First Energy’s grid with American Electric Power tripped
4:05 pm Sustained power surge on Ohio lines
4:09 pm Voltage sags as Ohio draws 2GW from Michigan
6
SCADA SystemsCAL-ISO Hacking
Hackers Victimize Cal-ISODan Morain, June 09, 2001
For at least 17 days at the height of the energy crisis, hackers mounted an attack on a computer system that is integral to the movement of electricity throughout California… The hackers' success, though apparently limited, brought to light lapses in computer security at the target of the cyber-attack, the California ISO, which oversees most of the state's massive electricity transmission grid.
Power Grid Incidents
7
Several prominent intelligence sources confirmed that there were a series of cyber attacks in Brazil: one north of Rio de Janeiro in January 2005 that affected three cities and tens of thousands of people, and another, much larger event beginning on Sept. 26, 2007. The attack in the state of Espirito Santo affected more than three million people in dozens of cities over a two-day period, causing major disruptions. In Vitoria, the world's largest iron ore producer had seven plants knocked offline, costing the company $7 million. It is not clear who did it or what the motive was.
Sanjay Goel, School of Business, UAlbany
200912 NOV; ONS, BrazilOperador nacional do Sistema Eletrico (ONS) is Brazil's national system operator responsible for controlling the transmission of electricity as well as the operation of generation facilities throughout the nation. On November 12th, a hacker gained access to its corporate network but stopped short of accessing its operational network.
8
TerrorismInternet Use
• Mobilization of public opinion / sympathy
• Spread of propaganda
• Solicitation of new recruits, donations
• Anonymous non-traceable communication
9
'Cyberwar' Emerges Amid Russia-Georgia Conflict
Georgia's recent conflict with Russia over the fate of two separatist provinces brought with it a first in international cyber-warfare, as Georgia faced a slew of Internet attacks.
Georgian government Web sites -- including the president's office, the parliament, and the foreign ministry -- were defaced with anti-Georgian or pro-Russian images. And Georgia's Internet system was crippled, as hackers manipulated computers to flood government, news, and information Web sites in a way that renders them useless.
Cyber WarfareRussia & Georgia
10
Munk Center in Canada shows that in less than 2 years, an electronic spying operation in China infiltrated at least 1,295 computers in 103 countries, including many belonging to embassies, foreign ministries and other government offices, as well as the Dalai Lama’s Tibetan exile centers in India, Brussels, London and New York.
Government InvolvementChinese Espionage on Tibetan Exiles
11
Cyber IntelligenceInternet: An Arena for Terrorists
12
• Data-mining works when– Search profile is well-defined
– Significant historical data for predictions
– Low cost of false alarms
• In espionage, counterintelligence, or terrorist plots– Uncertainty of what data to ignore or
pay attention to
– Attacks often hard to predict (little past data available)
– Avenues to hide involvement and communication
– False positives could lead to arrest of innocents and lost time on bad leads
Cyber Intelligence“Looking for a Needle in a Haystack”
We failed to stop 9/11 despite having critical intelligence
13
• No specific connection between real identity and internet aliases (can be multiple web identities)
• How is this done?– Anonymous web browsing, e.g. proxy servers– VoIP (e.g. Skype)– Private message boards– Chatrooms / IRC– Use of botnets (to send messages, relay, etc.)– Steganography with website / SPAM images
• Need intelligence techniques for assigning attribution (means, motives, and opportunity)
Cyber IntelligenceAttribution: Anonymity of the Internet
Has anonymity gone too far?
14
• Countless ways in which computer can be used to perform illegal activity
• Criminals leave behind traces that can be analyzed– Evidence in several media forms, e.g., text, audio, image,
video
• Multiple sources of data are needed to corroborate• We collect data from
• hacker blogs, websites, forums• Network (logs, SNMP, etc.) using darknets,
honeynets, and other devices
Cyber IntelligenceTracking Incidents
15
• Network Measures• Density (number of dyads connected to
each other)• Degree: Average number of connections• Path length: Average # arcs in shortest
path between two nodes• Clustering Coefficient: Measure of
grouping of nodes in graph• Centralization: Measure of cohesion in
graph• Node Measures
• Degree: Number of connections of a node• Betweenness: Critical link in the graph• Closeness: Average distance to other
nodes• Clique Count: # of cliques to which a node
belongs
Cyber IntelligenceSocial Network Analysis
Image Source: UMBC
HACKER NETWORK ANALYSIS ?
16
• A Simulated Signal Intelligence and Human Intelligence– Approximately 800 reports.– 8 month plot window.– 409 named entities.– 98 locations
Social Network AnalysisAlibaba Dataset
A 12 Member Terrorist Cell --- connected with the Ali Baba Network plans to “bake a cake” (build a bomb) which will be targeted to blow up a water treatment facility near London. The plot takes place from April to September of 2003
Robert Savell, School of Engineering, Dartmouith
17
• We are collecting data from targeted hacker forums/blogs/ websites• Project Grey Goose
• Natural Language Processing is being used for analyzing the data
• Process used for analyzing data– Develop seed list of relevant concepts in domain of interest and
cluster web pages– Develop concept graph for each cluster of documents, and use
concept co-occurrence distance and proximity filtering to reduce edge density;
– Identify related communities of concept terms within each resulting graph component.
– Manually assess each graph “community” and review the sets of
related pages for information of interest.
Open Source DataProximity of concepts
18
• Honeynets are networks of honeypots where all inbound and outbound traffic is collected– Multiple operating systems & applications
– Deploy services that closely match actual conditions in the organization
• Any attempt to contact to the network from outside is likely an attempt to breach its security
• Any outbound activity is likely evidence that a system has been compromised.
• Hacking tools can fingerprint honey pots/nets so they should be camouflaged
Network ForensicsHoneyNet
19
• Darknet is a portion of routed, allocated IP space where no active services or servers reside– Consists of a server that gathers packets & flows that enter the Darknet
• Blocks contain no active hosts, thus traffic must be caused by mis-configuration, backscatter from spoofed source addresses, or scanning from worms and other probing.
• Can be used in conjunction with flow collectors, backscatter detectors, sniffers and/or IDS boxes for further analysis
Network ForensicsDarkNet
20
• Creating a network of darknets across the globe with partners in Israel, Spain, Russia, India, United States
• Data can be collected in partner country or addresses can point to central data collection server in the United States
• Comparing changes in behavior across different darknets will help identification of activity patterns and identify malicious behavior
• Sharing honey net data further improves analysis quality
Network ForensicsDistributed Darknet: Creating a Global Telescope
21
• Improve ability to detect attacks and respond quickly (data collection and analysis)
• Collect global data• Develop techniques for improving analysis• Cooperative model based on shared intelligence
provides higher quality of alerts
Securing the InternetConclusions
ICFSInstituto de Ciencias Forenses y de la Seguridad
Próximamente: http://www.icfs.uam.es