how secure are your systems
TRANSCRIPT
How Secure are Your Systems? Quantitative
and Qualitative Assessment of Digital Security
Ilir Gashi, Vladimir Stankovic, Cagatay Turkay
Slide 2
Presenters
• All three presenters are from the Department of Computer Science, School of
Mathematics, Computer Science and Engineering (MaCSE) at City University London
• Ilir Gashi
Centre for Software Reliability (CSR)
W: http://www.city.ac.uk/people/academics/ilir-gashi
• Vladimir Stankovic
Centre for Software Reliability (CSR),
W: http://www.city.ac.uk/people/academics/vladimir-stankovic
• Cagatay Turkay
giCentre
W: http://www.city.ac.uk/people/academics/cagatay-turkay
Slide 3
The plan for tonight
• Start between 6.15pm-6.30pm (depending on audience numbers and late
arrivals). Aim to finish by 8.30pm.
• Part I - First 30-45mins
– Introduction to the concepts of Information Security
• Part II - 30mins:
– 10 minute video: TED talk on Stuxnet
– 10 minute group discussion on the video
– 10 minute plenary discussion
• Part III - Final 30-45mins:
– Overview of research on information security at the CSR, City University
London
• Questions invited throughout. We should be available at the end also for
further discussions
• The timings above (especially for parts II and III) may have to be revised
depending on the level of interaction
Slide 5
What is Information Security
• Various names:
– Information Security
– Information Assurance
– Cyber-security
• Protecting Information and Information Systems,
from misuse
– Misuse: Unauthorised access, use, disclosure,
disruption, modification, inspection, recording or
destruction
Slide 7
Key Security Concepts
• Confidentiality
– Preserving authorised restrictions on information access and
disclosure, including means for protecting personal privacy and
proprietary information
• Integrity
– Guarding against improper information modification or
destruction, including ensuring information nonrepudiation and
authenticity
• Availability
– Ensuring timely and reliable access to and use of information
Slide 8
Contextualising Information Security
• Information security draws upon the best practices and experiences from multiple domains
Slide 9
Some “truisms” about Information security
• There is No Such Thing as Absolute Security
– Given enough time, tools and skills hacker(s) can break through
any security measure
• Defense in Depth as Strategy
– There is no single magic bullet; a layered approach is always
necessary
• Security Through Obscurity is Not an Answer
• Security = Risk Management
– E.g. Spending more on securing an asset than the intrinsic value
of the asset is a waste of resources (e.g. buying a £1000 safe for
a £200 jewellery)
Slide 10
Some “truisms” about Information security (2)
• The Three Types of Security Control Are
Preventative, Detective and Responsive
– Good security will have a balance of all 3
• Complexity Is the Enemy of Security
– The more “moving parts”, the more likely it is that
some thing will break
• Fear, Uncertainty and Doubt (FUD) is not the right
way to justify security
– FUD needs to be replaced with a professional,
rational and balanced approach to security
Slide 11
Some “truisms” about Information security (3)
• Security is a Socio-technical problem
– People, Process, and Terminology Are All Needed to
Adequately Secure a System or Facility
• Open Disclosure of Vulnerabilities Is Good for
Security!
– Users should know about defects in the products they
purchase, just as they have the right to know about
automobile recalls because of defects
– Bottom line: if you uncover an obvious problem, raise
your hand and let someone who can do something
about it know
Slide 12
Vulnerabilities, Threats, Intrusions, Attacks
• We will go through definitions, concepts, examples, trends and
further reading resources for information security:
– Vulnerabilities
– Threats
– Intrusions
– Attacks
Slide 13
Sources
• We start with some definitions which will help us better frame the
many concepts that can be found in the security literature
• We will stick with the definitions adopted in the EU MAFTIA project:
– http://research.cs.ncl.ac.uk/cabernet/www.laas.research.ec.org/maftia/
• The deliverable D21 of MAFTIA is recommended for further reading:
– http://research.cs.ncl.ac.uk/cabernet/www.laas.research.ec.org/maftia/d
eliverables/D21.pdf
Slide 14
Some definitions - MAFTIA
• Fault: adjudged or hypothesised cause of an error
• Error: that part of the system state that may lead to failure
• Failure: delivered service deviates from implementing the system function
• Attack: a malicious interaction fault, through which an attacker aims to
deliberately violate one or more security properties; an intrusion attempt.
• Vulnerability: a fault created during development of the system, or during
operation, that could be exploited to create an intrusion
• Intrusion: a malicious, externally-induced fault resulting from an attack that has
been successful in exploiting a vulnerability
Slide 17
Summary of MAFTIA terminology
• Very useful terminology to conceptualise the security “threat” cycle
• However, as was the purpose of the MAFTIA project, the definitions tend to
come from an intrusion tolerance viewpoint
• A good source for definitions of security concepts can be found in the various
800-series of NIST Special Publications:
– http://csrc.nist.gov/publications/PubsSPs.html
Slide 19
Vulnerability lifecycle
• Another useful way of conceptualising the security threats is to look at the
lifecycle of a vulnerability
• A very good resource for this is the PhD thesis of Stefan Frei, from IBM
Zurich. A shorter version of this work is given in the following publication :
– http://weis09.infosecon.net/files/103/paper103.pdf
• Looking at the vulnerability lifecycle in this way gives us an insight to the
different types and stages of risks that a system may have
• It also gives us an insight on the economic aspects of vulnerability
disclosure and the different types of incentives that effects the decision
making of both attackers and defenders
Slide 22
Economics of security
• We touched upon earlier in an area of research that has received a lot of
attention in the last decade, namely the economics of computer security
• This research was motivated by the realisation that security system failures were
caused at times as much due to wrong incentives as they were from technical
errors
• One of the pioneers of this research is Prof Ross Anderson and a good starting
point on further reading on this topic is from his webpage:
• http://www.cl.cam.ac.uk/~rja14/econsec.html
Slide 23
Types of threats
• The most publicised threats to security are:
• Malicious software – commonly referred to as malware
• Intruders – commonly referred to as hackers or crackers
• In the next few slides we will look into:
• Types and examples of malware;
• Types, examples and techniques used by intruders:
• We will also discuss the possible motives of the attackers, and the capabilities
they may need to launch these attacks successfully
Slide 24
Malware (malicious software)
NIST* defines malware as:
“a program that is inserted into a system, usually
covertly, with the intent of compromising the
confidentiality, integrity, or availability of the
victim’s data, applications, or operating system
or otherwise annoying or disrupting the victim.”
* http://csrc.nist.gov/publications/nistpubs/800-83/SP800-83.pdf
Slide 26
Attacker motivations
• Financial
• Commonly from Criminals motivated by financial gain
• Political
• Nation state actors (defence or attack; activist/hacktivist groups (e.g. Anonymous
group))
• Military / Strategy
• Nation state actors (defence or attack (Stuxnet, Flame))
• Psychological
• Nation state attackers seeking to cause confusion and doubt in population etc.
Slide 27
Types of attacks
• As mentioned previously, attacks will seek to exploit a vulnerability in a system
• They may use malware as launching pads or for carrying out an exploit
• They may come from inside or outside the system
• They may be motivated by a variety of reasons
• Other examples of attacks (list is not exhaustive):
• Denial of service
• Distributed denial service
• Buffer over flow attacks
• Etc.
Slide 28
Other types of threats
• Organisational:
• Wrong or incomplete operating procedures
• Wrong or incomplete policy requirements, design or implementation
• Human error in carrying out a possibly correct policy or correct procedures
• Threats from misconfiguration of protection tools
• Supply chain threats from use of “software of uncertain pedigree” (COTS, OTS etc)
• Many of these threats also apply to other environments such as “cloud” or “mobile”
though with possibly different probabilities of being exploited and possibly different
“risk ownership”
Slide 29
Where to find more information
• Many security vendors publish “threat reports” frequently:
• See example from Symantec or IBM:
• http://www.symantec.com/threatreport/
• http://www-03.ibm.com/security/xforce/threats.html
• Though one has to be cautious when interpreting the findings of these reports and look
carefully at the data collection methods. Caution is especially necessary in cases where
cost estimates are reported:
• See a recent report co-authored by Prof Ross Anderson on this topic:
• http://weis2012.econinfosec.org/papers/Anderson_WEIS2012.pdf
Slide 30
Where to find more information (2)
• Other, independent, sources of evidence on threats and vulnerabilities:
• http://cve.mitre.org/ (which also contains links to other endeavours within the
Mitre corporation for enumerating attacks, weaknesses, platforms etc.)
• http://nvd.nist.gov/
• The NIST vulnerability database
• http://www.us-cert.gov/
• The US Department of Homeland Security's Computer Emergency Readiness
Team (US-CERT)
• http://www.ukcert.org.uk/
• UK CERT (mainly a portal with links to other worldwide CERTs)
Slide 31
In conclusion
• What advice can we give to an information systems professional?
• Know your system and your environment
• Be clear about what threats are relevant to your environment and what are not:
this help you make informed decisions on security spending
• Careful from marketing hype and be clear about the data collection and analysis
methods on threat reports – and whether the findings apply to your organisation
• Understand what the reports contain and how this should influence the decisions
you make on the security policy and security budget
• Be wary of risks in the supply chain and risks from use and deployment of different
environments (such as cloud and mobile)
Slide 32
References
• See slides for additional references
• Text-1 - Chapters 1 and 2
• Text-2 - Chapters 6, 7 and 8
• Check out TED talks from the series on hackers:
– http://www.ted.com/playlists/10/who_are_the_hackers.html
– These talks will give you a wider perspective of the types of attacks that are
being performed, by whom, and what the motivations may be
• Text-1: Information Security Principles and Practices by Mark Merkow and James Breithaupt,
Prentice Hall, 2006
• Text-2: Computer security : principles and practice, William Stallings, Lawrie Brown (2nd Edition
(International)), Pearson
Slide 35
Activity
• (10 mins) We will have a look at the following TED talk on Stuxnet:http://www.ted.com/talks/ralph_langner_cracking_stuxnet_a_21st_century_cyberweapon.html
• (10mins) There have been many articles published about Stuxnet
and its authorship has been attributed to many different sources
(though common consensus now is that it was most likely the work
of the US intelligence). In groups discuss who the attack sources
may be, and in particular for each discuss:
– What would be the motivation (economical, political, strategy, psychological)
– What level of skill would be needed to develop such a malware
– What are the short and long term implications of Stuxnet. Who is more likely to
be at risk in the long term from its disclosure
(think about the targets of Stuxnet)
• (10 mins) Plenary discussion
Slide 37
Acknowledgments
Work we will summarise next has been done with many colleagues at CSR:
Lorenzo Strigini, Eugenio Alberdi, Robin Bloomfield, Bev Littlewood, Peter Popov, Peter
Bishop, Andrey Povyakalo, Kizito Salako, Robert Stroud, David Wright, Katerina Netkachova,
Oleksandr Netkachov and several ex-colleagues and collaborators
Slide 38
Outline of presentation
– CSR and the experience we bring to security topics
– Some examples of research and applied work
Slide 39
Background
– Centre for Software Reliability (www.csr.city.ac.uk)
– Founded in 1983 to deal with problems surrounding the [un]reliability of
software
– Quickly expanded into a wider systems viewpoint, dependability/security of
socio-technical systems
– Distinctive features
– Emphasis on rigorous assessment (esp. probabilistic)
– Developed models for empirical assessment as well as for insight
– Reliability growth, very high reliability, fault tolerance
– Interdisciplinary approach
– Extensive work on redundancy, diversity, defence in depth (more on this
later in the presentation)
– Work with industry, e.g. long relationship with nuclear safety research;
collaborations with Adelard, a safety consultancy
– All this affects what we bring to security research
Slide 40
Some projects with security contents
www.csr.city.ac.uk/projects
• EU: PDCS (1989-1995) (Predictably Dependable Computer Systems)
• EPSRC: DIRC (2000-2006) (Interdisciplinary Research Collaboration in Dependability of
Computer-Based Systems)
• EU: ReSIST (2006-2008) (Resilience for Survivability in Information Society
Technologies(IST))
• EU: IRRIIS (2006-2009) (Integrated Risk Reduction of Information-Based Infrastructure
Systems)
• EPSRC: INDEED (2006-2010) (Interdisciplinary Design and Evaluation of Dependability)
• EU: AMBER (2008-2009) (Assessing, Measuring, and Benchmarking Resilience)
• PIA:FARA (2009 - 2010) (Probabilistic Interdependency Analysis: framework, data analysis
and on-line risk assessment)
• EU: AFTER (2012-2014) (A Framework for electrical power systems vulnerability
identification, defence and restoration)
• EU: SeSaMo (2012-2015) (Security and Safety Modelling)
• Leverhulme: UNcertainty and COnfidence in DEcision making
• EPSRC/CPNI/Research Institute in Trustworthy Industrial Control Systems: CEDRICS
(2014-1017)
• EPSRC: D3S (2015-2018) (Diversity and Defence in Depth for Security)
Slide 41
Some themes and topics
– The role of quantitative reasoning in security– Necessary basis for decision making
– Security as system issue– computing as part of a whole with people and physical systems– socio-technical systems: people’s behaviour in building, operating, working
as part of systems – Empirical study of threats
– e.g., data collection via Honeynets– Fault tolerance, diversity, defence in depth
– theory and modelling for decision support– empirical measurement, e.g. antivirus, biometrics
– Convergence of security, safety and reliability concerns– embedded systems (SeSaMo, CEDRICS)
– Publications at: www.csr.city.ac.uk/security
Slide 42
Fault tolerance and diversity for security
- To improve reliability and safety, it is normal to
– Replicate systems and information
– Use diverse systems to achieve the same purpose
• Less likely to fail in same situation
– The approach is valuable in IT security as well
• Vulnerability of "monoculture" to "infection"
• But curiously controversial
• Need to match techniques to threats, make trade-offs explicit
– Our contributions include modelling, empirical assessment
• How to assess potential benefit / risk of multiple layers of defence
• Experimental measurement
• Further details on CSR’s work in this area: www.csr.city.ac.uk/diversity
Slide 43
Some examples of our research
- In what follows we will give you a few examples on the CSR work on
security (the list is not exhaustive):
- Diversity with AntiVirus products and Operating systems
- Interoperability of fingerprint biometric systems
- Resilience and security/dependability for critical infrastructures
- Security review of European Railway Traffic Management System
Slide 44Empirical assessment of the effectiveness of
combining antivirus engines
Motivations
• Decision making in security – often driven by subjective anecdotal evidence and
marketing hype
• Objective empirical evidence is needed for sound decision making
– otherwise difficult to evaluate and objectively justify decisions
• A study on the potential benefits of using diversity with an important kind of off-the-
shelf software: AntiVirus products
– used and advertised in commercial products, but without evaluating effectiveness
• We have performed several studies in collaborations with our research partners at
Eurecom Institute in France, Symantec Research and University of Maryland.
• Here we describe the first study – conducted in 2008 :
– 32 AntiVirus engines hosted by the VirusTotal site
• We concentrated on 1 component of AVs – signature-based engines
• Our goal: evaluation of diversity effectiveness, not ranking of AVs!
– with 1599 malware samples collected from a distributed honeypot platform
– over a six month period between February and August 2008
Slide 45
Experimental setup and architecture
Anubis
Gate
way
SensorS
SH
Sensor
Malware
Sensor
Malware
Slide 46
A contour plot of the AV (x-axis) failure rates (z-axis, represented
by the intensity of the colour in the plot) over the malware (y-axis)
• Malware samples submitted to the AVs repeatedly (30+ days)
• Input to the analysis – triplet {Malwarei, AVj, Dayk}
Slide 47
Single AV Product Results - Regression
• Large number of AVs fluctuated
in their detection capability.
• In many cases, the AVs
regressed in their detection
decisions:• Detected the malware at first, and
then failed to detect the malware
at a later date
• A few of the AVs in the Top 10
for their detection capability are
among the AVs that regressed
Slide 48
Regression example
• Two AVs (AV-20 and AV-8) detect the malware on most days but not
every day.
• Both of the AVs fail to detect the same malware on the same day
only once: in other days when one AV fails the other does not.
• This would indicate gains from employing diverse AV engines
Slide 49
Main results of initial analysis
– Main findings from this data set:
• Considerable variability in detection capability
• None of the AVs on their own achieved 100% detection rate.
• Significant improvements in detection rates from using:
– 1-out-of-2 diverse pairs of engines
» 91 pairs, 18% “perfect” (zero failure rate)
– high levels of diversity (1oo3, 1oo4 etc)
» 163 pairs, ~33% better than the best single AV
– Further research performed and reported in:
• http://openaccess.city.ac.uk/1526/
• http://openaccess.city.ac.uk/2716/
• http://openaccess.city.ac.uk/2338/
Slide 50
Operating systems’ vulnerability diversity
• We have also looked at the extent to which operating systems are
diverse in the vulnerabilities they contain
• We used data from the NIST National Vulnerability Database to
empirically evaluate the vulnerability diversity
• We found a high degree of diversity amongst certain families of
operating systems
• This might have implications on the optimal choice of diverse
Operating systems that administrators may wish to use (especially
when deciding on which OSs to use for web server or database
server deployments etc.)
• Details:
– http://openaccess.city.ac.uk/2127/
Slide 52
Biometric fingerprint system interoperability
• In collaboration with colleagues from University of West Virginia, we
performed analysis of a large scale empirical study on the effects of
interoperability with biometric fingerprint systems.
• In our study, the fingerprint images from:
– 494 participants were collected with
– 4 different fingerprint image scanners, the quality of the images
assessed by
– 2 different image quality algorithms, and matching scores
calculated with
– 3 different matching algorithms
• The results were then further categorised by:
– Age, Gender, Height, Weight, Ethnicity etc.
• To allow us to make empirically-supported recommendations for
biometrics systems deployment
Slide 53
What are Biometrics?
• A means of authentication/identification of an individual based on
unique physical characteristics/traits.
• Static Biometrics
– Something that the individual IS
• Dynamic Biometrics
– Something that the individual DOES
• Passwords are something the individual knows.
Slide 54
Example Deployment
• Consider a Border Agency scenario, that requires foreign nationals to
verify their identity before entering the country.
– Enrolment
• As part of the visa application process, the individual must
present their unique biometric information (i.e. fingerprint image).
• This biometric information is stored in the system as a ‘template’ for the individual.
– Verification
• Upon attempting to enter the country, the individual presents their visa as
identification.
• To verify this identification, the fingerprint image is recaptured.
– The feature set is compared with the template to obtain a match similarity
score.
• Based on a set score threshold, the individual is either allowed/denied access to
enter the country.
Slide 55
Biometric System Performance
• Performance is based on the overlap of imposter and genuine scores.
– False Accept Rate = Imposter Accepted
– False Reject Rate = Genuine Rejected
– Equal Error Rate where (FAR == FRR)
• How should the acceptance Threshold be set?
– Depends on the operating environment.
• Types of operating environment.
– Low FAR – high security / military.
– Low EER – civilian applications.
– Low FRR – forensics or low-cost for false accept.
Slide 56
Visualisation – how can it help a decision maker
• We explore how (some initial) data visualization techniques may help a decision maker with
– deployment of biometric systems which
– balance the trade-offs between cost and failure rates
– when combining the different components (e.g., fingerprint image capture devices, image quality algorithms and matchers)
– for a given environment.
• The dataset that the decision-maker would use in this scenario is multi-dimensional
– (e.g., different components (image quality, matchers, image capture devices) and multiple demographic aspects (age, gender, ethnicity, height, weight etc),
• Hence, a decision maker would benefit from data visualization techniques that allow for identification of trends in the dataset, anomalies and trade-offs in a fast and intuitive manner.
Slide 57
Three tasks to help us with visualisation
design• T1 : Visualize the ”distinctiveness” for each probe device, gallery
device and algorithm combination: the extent to which a
threshold value clearly separates the genuine scores from the
impostor scores.
• T2 : Visualize the overall inter-device, intra-device matching
performance and matching algorithm performance.
• T3 : Investigate dynamically through visualization the relation
between soft biometrics (age, gender, height, weight) and the
matching performance.
Slide 60
Short video of the visualisation tool
• Available from here:
– https://www.youtube.com/watch?v=CneKJwp2_n4
• Papers on biometrics:
– http://openaccess.city.ac.uk/3507/
– http://openaccess.city.ac.uk/4091/
Slide 61
Critical Infrastructure Interdependencies
• A key issue for achieving CI resilience and CI
protection
– risk of CI disturbances propagating across ‘dependencies’ links
• Complex phenomena, not well understood
Slide 62Modelling the Resilience and Interdependency of Critical
Infrastructures
• PIA:FARA Toolkit Prototype
• PIA Designer – qualitative analysis of possible interdependencies
• PIA Run-time support – quantitative, probabilistic model of interacting CIs
– a simulator of the model behaviour in the presence of failures of the modelled entities for the
chosen model parameterisation
• Supporting various analyses
• e.g. which parameters are most critical (highest impact on likelihood of cascade
failures)
• Implications of short- to mid- term trends, e.g. growing fraction of renewable energy
• Short term predictions, e.g. risk of a service disruption in the next ΔT.
• Test case and scenario generation
• E.g. for training or demonstration purposes
– Recently, expanded this work towards modelling of security aspects, e.g. modelling
adversaries behaviour/attacks and their impact on the modelled system.
• Work developed in various projects:
• EU-funded: IRRIIS (EU FP6 - IP), AFTER (EU FP7 - STREP)
• National funding: PIA-FARA (TSB, EPSRC), Cetifs (CPNI, EPSRC, TSB)
– Further details: http://openaccess.city.ac.uk/3091/ , http://openaccess.city.ac.uk/4361/
http://openaccess.city.ac.uk/4363/
Slide 63
Example: Security analysis of ERTMS
– the European Railway Traffic Management System (ERTMS) is a new standard
intended to create a seamless European railway system
– we were asked for a study of its security implications
• Upgrade of the UK railway signalling infrastructure over the next 30 years –
new system will be in operation for at least 50 years.
• Centralised control implies risk of an electronic attack bringing down the
entire network.
• Safety has always been paramount but now need to consider the security
implications.
• Need to understand how secure the ERTMS/ETCS specifications are and
whether the industry implementations may compromise the security
requirements.
• Investigate whether the technology is secure from external interference,
identify vulnerabilities and provide advice to mitigate any security flaws
– Main results of our study are still restricted, but a high level paper with results is
here: http://openaccess.city.ac.uk/1522/
Slide 64ERTMS system architecture
Balise
Control Centre
Controls and Indications
Interlocking
Train Detection
Track Occupancy
RBC
Movement Authority
Position Information
GSM-R
MSC/BSC
BTS
European Vital Computer
EVC
Balise antenna
DMI
Odometry
GSM-R Radio
Slide 65
Further information and resources
• InfoSec Europe is an annual industry event that takes place in London. This
year it will be in Kensington Olympia:
http://www.infosecurityeurope.com/
• We run two MSc courses on information Security at City:
– MSc in CyberSecurity
• http://www.city.ac.uk/courses/postgraduate/cyber-security
– MSc in Management of Information Security and Risk (MISR)
• http://www.city.ac.uk/courses/postgraduate/management-
information-security-risk-msc
• All of MISR modules can be taken individually as CPD modules
• We are also happy to discuss further any possibilities for research
collaborations. For example via Knowledge Transfer Partnerships:
– http://ktp.innovateuk.org/
Slide 66
D3S project – recruiting a research associate
• We are starting with a new Research Project on defence in depth for
security:
– http://www.city.ac.uk/news/2015/march/researchers-at-citys-centre-for-
software-reliability-are-the-recipients-of-a-563,890
• We will be recruiting a research associate with a 3 year full-time contract
• There may also be a possibility of studying towards a PhD in parallel to this
employment
• If you are interested, please get in touch: