1.2

Page 3: Authentication

• Authentication is the binding of an identity to a subject (a user or an entity).

• A user or entity is often required to authenticate itself to a computer system.

• When using Internet banking, authentication by the bank site is necessary.

• When using email, a password linked to the username or account name is required for authentication.

Page 4: A Digital User Authentication Model

• NIST SP 800-63-3 defines a general model for user authentication.

• It involves two steps: user registration with the system and authentication.

Page 5: Registration, credential issuance, and maintenance

• Registration involves identity proofing, subscriber/registration authority, and user registration claimant.

• Credential issuance includes authenticated confirmation assertion and token/credential service.

• Maintenance involves relying party, authentication session authority, and validation provider.

Page 6: Means of Authentication

• A subject must provide information to confirm its identity.

• Information can be something the individual knows (password, PIN, answers to questions), possesses (token), is (static biometrics), or does (dynamic biometrics).

Page 7: Multifactor authentication

• Multifactor authentication uses more than one authentication means.

• The strength of the system is determined by the number of factors incorporated.

• Two factors are considered stronger than one, three factors are stronger than two.

Page 8: Password-based Authentication

• Password-based authentication is a simple and common method.

• Despite security vulnerabilities, passwords are commonly used for authentication.

• A password is information associated with an entity that confirms its identity.

Page 9: How do password systems work?

• Account registration can be allocated/set by an administrator.

• Passwords should be changed in this case.

• Double entry of passwords and confirmation through email are common.

• Authentication and reset/update/recover are part of the process.

Page 10: Password Authentication

• The user supplies an identity and a password.

• The server checks the supplied information.

• If the password matches, the user's identity is authenticated; otherwise, the password is rejected.

Page 11: False positives and negatives

• False positives and negatives are errors in authentication.

• False positives occur when a match is made but shouldn't have.

• False negatives occur when a match is not made but should have.

Page 12: False positives and negatives (continued)

• False acceptance rate (FAR) is the proportion of authentication attempts resulting in false acceptances.

• False positive effects differ in the context of "raising an alarm" from the context of "making a match".

Page 13: False positives and negatives (continued)

• False rejection rate (FRR) is the proportion of authentication attempts resulting in false rejections.

Page 14: FAR and FRR

• FAR and FRR are possible in authentication mechanisms with tolerance in matching.

• Initially, with no tolerance, these rates default to zero.

• FAR becomes relevant when storing transformed passwords.

• Cryptographic hash functions are used for this purpose.

Page 15: Threats against password systems

• Password guessing, exposure, login Trojan programs, and poor passwords are threats.

• Common attacks include dictionary attacks, brute force attacks, and hybrid attacks.

• Compromise of the password file can occur online or offline.

Page 16: Password guessing

• It is always possible to attempt to guess a password online.

• Guessing the password of a user on Capa by trying to login as that user is a common method.

• Depending on the authentication mechanism, the data sent to the server may not be the actual password.

Page 17: Password exposure

• Passwords can be seen by eavesdroppers when typed.

• Writing passwords down or sharing them with others is a security risk.

• Trust assumptions are critical in security.

Page 18: Login Trojan Horses

• Login Trojan Horses produce genuine login screens but capture passwords.

• The captured information is stored for malicious purposes.

• Protection lies in not installing the Trojan Horse in the first place.

Page 19: Poor password

• Users often choose simple passwords for ease of remembering.

• Many systems enforce restrictions on the passwords allowed.

Page 20: Poor passwords → Dictionary attacks

• Password requirements may allow dictionary words.

• Dictionary attacks use sets of common words to try as passwords.

• This attack is fast but may not always succeed.

Page 21: Tailored dictionary attacks

• Dictionaries can be specific to a theme (e.g., car brands, sports teams).

• Users may use personal information like birthdates or family names for passwords.

Page 22: Brute force

• All password systems are vulnerable to guessing the correct password.

• Brute force attacks involve trying every possible password.

• Changing passwords regularly makes it harder to guess.

Page 23: Choosing secure passwords

• The time to test the correct password depends on the size of the password space and the number of passwords that can be tested per second.

• Randomly generated passwords with a large character set are secure but hard to remember.

• Using an arbitrary password over a larger character set increases security.

Page 24: Password entropy

• Entropy is related to information content, randomness, and uncertainty.

• Entropy is often measured in bits.

• The entropy for N equally likely options is log2N.

Page 25: Example: compute password's entropy

• The password is generated with specific criteria.

• The password's entropy needs to be computed.

Page 26: Example: compute password's entropy

• The password space is computed by determining the number of possible passwords for each character group.

• The total number of possible passwords is calculated.

• The entropy is determined using the formula log2N.

Page 27: Trying to improve passwords

• Pronounceable passwords make it easier to remember passwords.

• Using pass-phrases with intentional misspellings, odd capitalizations, and symbol replacements can enhance password security.

Page 28: Hybrid attacks

• Hybrid attacks combine elements of dictionary attacks and brute force attacks.

• Variants of words tested in a dictionary are used.

• Characters can be replaced with numbers or symbols.

Page 29: Personal phrase based: Helping memory?

• Choosing a phrase and taking the first letter from each word can help with password memorization.

• Using well-known phrases is not recommended due to the risk of dictionary attacks.

Page 30: Protective mechanisms

• Tracking incorrect password attempts can help identify potential intruders.

• Limiting the number of guesses per connection attempt or locking the account after a threshold is exceeded can deter attackers.

• Slowly processing passwords can slow down attackers.

Page 31: "Online" versus "Offline" guessing

• Online guessing faces restrictions on the number of attempts, while offline attacks do not.

• Offline attacks can occur if an intruder accesses the password file or intercepts the transmission of a password.

• Communication security and cryptography are covered in other courses.

Page 32

• The distinction between "online" and "offline" guessing is not as important as the restriction on the number of guesses.

• Attackers will operate differently based on the guessing restrictions.

Page 33: Rules for password systems

• Examples of rules for password systems are provided.

• Passwords should be changed regularly, have a minimum length, and include a mix of characters.

• Dictionary words should be avoided, and previous passwords should not be reused.

• Failed logon attempts can result in temporary account lockouts.

Page 34: UOW password rules

• The University of Wollongong has specific password rules.

• Passwords must meet certain length and character requirements.

• Personal information and compromised passwords are not allowed.

Page 35: Protecting passwords

• Password repositories must be well protected.

• In UNIX, only hashes of passwords are stored.

• Hashing makes it computationally infeasible to find the associated password.

Page 36: What is hashing?

• Hashing is a procedure used for message integrity and indexing.

• Hashing reduces the computational overhead of digital signatures.

• The hash of a message is a fixed-length fingerprint of the data block.

Page 37: Hash Functions

• Hash functions transform data into a fixed short length.

• The hash value of a message can be efficiently computed.

• The output is referred to as the hash value or message digest.

Page 39: Cryptographic Hash Functions

• Cryptographic hash functions have additional properties.

• One-way or pre-image resistance makes it computationally infeasible to find the original message from the hash.

• Collision resistance makes it computationally infeasible to find different messages with the same hash.

Page 41

• MD5 and SHA-1 are hash algorithms.

  • MD5 produces a 128-bit message digest.

  • SHA-1 produces a 160-bit message digest and uses the design approach used in MD5.

• Both MD5 and SHA-1 are broken with respect to collisions.

• Collision resistance is harder to achieve than pre-image resistance.

Page 42

• Additional sources on hash function security and encryption:

  • Wikipedia page on hash function security summary

  • NIST policy on hash functions

  • Street Directory article on MD5, SHA-1, and SHA-2

  • FreeCodeCamp article on secure encryption hash

  • Jscrambler blog post on hashing algorithms

Page 43

• Passwords are not directly stored in UNIX, only their hash is stored.

• The password file is encrypted using a system-known password.

• The password is inputted during system boot.

• Salting is used to further protect the hash.

Page 44

• Password salting involves using a random or pseudo-random value called a salt.

• The hash of the salt and password combination is stored.

• The salt value is also stored.

Page 45

• The salt prevents duplicate passwords from being visible in the password file.

• Different salt values for the same password result in different hashed passwords.

• Salting increases the difficulty of offline dictionary attacks.

• A salt of length b bits increases the number of possible passwords by a factor of 2^b.

• It becomes nearly impossible to determine if a person has used the same password on multiple systems.

Page 48

• Hash collisions can lead to false positives in password acceptance.

• The likelihood of collisions is very low.

Page 49

• Early versions of UNIX stored user IDs and transformed passwords in the /etc/passwd file.

• The /etc/passwd file contained user ID, encrypted password, home directory, and default shell information.

Page 50

• The general format of the passwd file includes fields for username, transformed password, UID, GID, full name, home directory, and shell.

Page 51

• UNIX split the passwd file information into two files: passwd and shadow.

• The passwd file contains everything except the protected passwords.

• The shadow file contains the transformed passwords and is only accessible to the root user.

Page 52

• Storing password hashes in shadow files is the preferred method.

• The /etc/passwd file should not store password hashes.

• The /etc/passwd and /etc/shadow files contain corresponding username and transformed password pairs.

Page 53

• Shadow files have fields for username, transformed password, last password change date, minimum and maximum password change days, warning days, expiration days, and disable days.

Page 54

• Using shadow files is safer than before, but there are still vulnerabilities.

• Attackers need a valid user ID and password to break into a system.

• Valid user IDs can be obtained from the /etc/passwd file.

• Password guessing attacks can be launched.

• Shadow files require root access, but there were attacks that allowed acquiring the shadow file without root access.

• Certain servers like imapd and telnet were guilty of dumping core with the shadow file in a user-readable format.

Rainbow Tables

• Pre-computation can speed up an attack

• Rainbow tables are lookup tables of passwords and corresponding hash values

• Extended versions use hashing and reduction functions for smaller lookup tables

Reduction Functions

• Map hash output space back into password space

• Table construction involves hashing and reducing passwords in sequences called hash chains

Table Lookup

• Check hash value in the table

• If not found, reduce and hash until it appears

• Found value indicates the password to start from

• Hash, reduce, and hash until the original hash value is reached

• The preceding password is the desired one

One-time Passwords

• User and system have a list of valid passwords, each valid only once

• Immune to eavesdropping and leaked passwords reveal no information

• Problems include the number of passwords to be shared and stored, significant initial costs, and increased storage for the server

Lamport's One-time Password

• User remembers a password, server stores username, counter, and hash value

• Authentication protocol involves exchanging information and checking hash values

• Server updates information after successful authentication

Alice and Bob

• Common names used to represent participants in cryptographic protocols

• Other cryptographic identities include Eve, Mel, Oscar, Peggy, Victor, etc.

Token-based Authentication

• Tokens are objects used for user authentication

• Two types: memory cards and smart cards

Memory Cards

• Can store but not process data

• Examples include bank cards and hotel room cards

• Authentication requires both the memory card and a password or PIN

• Adversary needs physical possession and knowledge of the PIN

Smart Cards

• Include an embedded microprocessor

• Can have manual or electronic interfaces

• Provide means for user authentication through static or dynamic password generation or challenge-response

• Contains ROM, EEPROM, and RAM memory types

Biometric Authentication

• Should be used as a component of a multi-factor authentication system

• Biometrics are not private and are used to make attacks more difficult

• Face recognition, handwriting, fingerprints, iris codes, voice recognition, and DNA are types of biometrics

• Iris codes are considered the most robust, but still have some problems

provides a rough indication of the cost and accuracy of biometric measures.