Title: Security Engineering
1Security Engineering
- (overview)
- John.Giacomoni_at_colorado.edu
- 2002.01.14
2Purpose of Security Engineering
- Resource Protection
- Confidentiality
- Integrity
- Availability
3Mechanisms
- Access Control
- Authentication
- Protocols
- Mediation
- Audit
4Mechanisms(Access Control)
- Restricting Access to Named Subjects
5Mechanisms(Authentication)
- Ensuring Authenticity of Subject
- (is the participant the subject it claims to be)
6Mechanisms(Protocols)
- Information Exchange Mechanisms
7Mechanisms(Mediation)
- Mechanisms Used to Grant Resources (objects) to
Subjects
8Mechanisms(Audit)
- Usage Reports
- Forensic Analysis
- Intrusion Detection
9Toolbag
- Cryptography
- Symmetric (Shared Secret)
- Asymmetric (Public/Private Keys)
- Digital Signatures
- Secure Hashes
10Whats Good with Cryptography?
- Strength based on Key
- Algorithm/Protocol can be Publicly Accessible
- Foundation in Clean Mathematics
- Algorithms Publically Reviewed and Accepted are
Rarely Broken In Production Systems.
11Results of Good Cryptography
- Seen as a Panacea
- Rarely Directly Attacked
12Mechanism Failures(Access Control)
- Collusion
- Manager Gives Personal Keys to Repair Technician
- Naming Attacks
- DNS Cache Poisoning
- Certificate Authority Database Corruption
13Mechanism Failures(Authentication)
- Collusion
- Manager Gives Personal Keys to Repair Technician
- Identity Theft
- Stolen Authentication Secrets/Tokens
14Mechanism Failures(Protocols)
- Eavesdropping
- Unencrypted Sniffing
- Reflection
- MIG/Man-in-the-Middle
- Logic Errors
15Mechanism Failures(Mediation)
- Authentication/Naming
- Corruption of Trusted Computing Base
- Logic Errors
16Mechanism Failures(Audit)
- Audit Trail Altered
- Audit File NOT Append-Only
- Audit Trail Overloaded
- Overloading Audit Log Partition
- Access Control/Authentication Failures
17Advanced Mechanisms
- Multilevel Security
- Multilateral Security
18Multilevel Security
- Bell-LaPadula
- Read Down
- Write Up
- Confidentiality
- Biba
- Read Up
- Write Down
- Integrity
19Multilevel SecurityWhat goes wrong?
- Data Floats up and Stays Up
- Composability
- The Cascade Problem
- Covert Channels
- Viruses
- Polyinstantiation
20Multilateral Security
- Compartmentalization
- Parallel Compartments in Isolation
- Chinese Wall
- All Compartments Available Until a Choice is Made
- BMA
- Complicated (See me for details)
21Multilateral SecurityWhat Goes Wrong?
- Inference Control
- Recomposition of Data
- Generic Approaches Limited in Scope
22Things Which Go Wrong(In General)
- Emergent Properties
- Environmental Creep
- People
- Economic Incentives
23Emergent Properties
- Bugs
- Subliminal Channels
- The Cascade Problem
- Recomposition of Data
24Environmental Creep
- Changing Environment Without Reevaluating
Specifications. - Old TCP Protocols (telnet, ftp, etc) on Global
Internet - ATMs, Banks -gt Convenience Stores
25People(who are you protecting against?)
- Skill Non/Moderately/Highly
- Motivation Idle/Moderately/Highly
- Resources Low/Medium/High/Intelligence Agency
- Location Inside/Outside
26Economic Incentives
- Information insecurity is at least as much due
to perverse incentives. - Network Externalities
- Moral Hazard
- Liability Dumping (Tragedy of the Commons)
27What are we Left With?
28Programming Satans Computer
- programing a computer which gives answers
which are subtly and maliciously wrong at the
most inconvenient possible moment.
29Is the Security Engineer S Out Of Luck?
- NO
- Solid Foundation Technologies
- Increasing Awareness
- Though not Enough
- Case Studies
30Case Studies
- Military Applications and Intelligence Agencies
- Identify Friend/Foe
- Radio Communications
- Industry
- ATMs
- Pay-TV
31Case Study(Pay-TV)
- Discard Messages Addressed to System
- Then Cancel Subscription (Yay Free TV)
- Freeze EEPROM by Removing Programming Voltage
- Eventually Bypassing SmartCard Security Features
with Probing Stations etc - 10,000 for a second hand one
32Risk Management
- The Systematic Evaluation of Risk in the Scope of
the Encompassing Business Model - 10,000,000 Raw Gross at 1 Loss 9,900,000
- 10,000,000 Raw Gross at .5 Loss 9,950,000
- 20,000,000 Raw Gross at 2 Loss 19,600,000
- (Increased Loss Acceptable in Face of Greater
Realized Gross)
33Security Requirements Engineering
- Risk Management
- Security Policy
- Security Target
- Security Policy and Target Engineering is not
Intrinsically Different from the Normal
Engineering Process
34Evaluation Assurance
- DoD Rainbow Series
- US Gov. Specified and Tested
- Gov. Carries Expense of Evaluation
- Common Criterion
- Preexisting Protection Profiles
- Independent Evaluators
- Producer Pays for Evaluation
- Supposed to be Cheaper/Fairer
35Beware 3rd Party Evaluations
- Who is Evaluating the Evaluators?
- (Economic Incentives)
36Summary
- Traditional Approaches Fail
- Security Needs to be Pervasive
- A Systems Approach is Required
- Technical
- Socio-Economic
37FYII have Skipped the Following Topics
- Distributed Systems
- Monitoring Systems
- Alarms/IDS
- Zero Failure Tolerance Systems
- Nuclear Command and Control
- Security Printing/Seals
- Biometrics
- Tamper Resistance/Detection
- Emissions Security (EMF/Tempest)
- Electronic Information Security
- Telecom Systems
- Network Attack and Defense
38For More Information
- Security Engineering by Ross Anderson
- Come Talk to Me )
39QA Time )
- (Look a plane ltrun awaygt)