Transcript User Authentication

12. Protection/Security Interface
12.1 Security Threats
– Types of Damage
– Vulnerable Resources
– Types of Attacks
12.2 Functions of a Protection System
12.3 User Authentication
– Approaches to Authentication
– Passwords
12.4 Secure Communication
– Principles of Cryptography
– Secret-Key Cryptosystems
– Public-Key Cryptosystems
Operating Systems
1
Security Threats
• Types of damage
1. Information Disclosure
• theft
2. Information Destruction
• possible without disclosure
3. Unauthorized Use of services
• install SW without license, pirated copies (theft)
• use fake ID/password to use online service
4. Denial of Service
• difficult to quantify
• Vulnerable resources
– Hardware (CPU, memory, communications, devices)
– Software (files, processes, VM)
Operating Systems
2
Types of Attacks
• Browsing of Information
– Unauthorized search for residual information
– Unused memory and disk space is generally not deleted
– Typically done by a user who is already inside
• Information leaking
– A trusted service leaks confidential/secret information
(Confinement Problem)
Operating Systems
3
Types of Attacks
• Trojan Horse
– Greek mythology—the siege of Troy
• wooden horse = “present” by Greeks
• soldiers hidden inside
• Trojans pulled the horse into the city
• soldiers opened the gates for the Greeks,
causing the destruction of Troy
• Attack: trusting user accepts a “present”, e.g. a free program,
that causes damage (don’t open email attachments)
• Trap door
– an undocumented feature
– inserted on purpose or as a flaw to enter later
Operating Systems
4
Types of Attacks
• Viruses
– Designed to replicate themselves
• removable storage media, email, file transfer
– Intended to cause damage
– Need a host program
• attach to and modify host
• execute as part of host
– Virus detection
• check program length
(but virus can hide or compress program)
• check for virus “signature”—bit pattern used by
virus to mark already infected program (viruses use
encryption)
Operating Systems
5
Types of Attacks
• Worms
– Intended to cause damage
– Exploit some system weakness to replicate
– No host needed
• Example: Robert Morris Internet Worm (Nov 2, 1988)
– Excessive replication caused major havoc on the internet
(denial of service)
– 3 separate attacks:
• rsh: Spawn process on remote machine without pw
(using a list of “trusted” systems)
• sendmail: Exploited an error that allows a message
to send itself and start
• finger: Buffer overflow not checked – major
weakness to take over the system
Operating Systems
6
Types of Attacks
• Buffer Overflow:
Example:
– foo calls finger
– Attack hijacks return
address by supplying a
parameter that is
longer than the buffer
(overflows)
– When finger terminates,
control goes to a place
set by the attack and
is not returned to foo.
Operating Systems
7
Types of Attacks
• Remote execution
– Service to upload and start code on remote machine
– Mobile agent: may migrate among machines
– Like worm but legitimate migration
– Must be designed carefully to prevent abuse of
privileges of remote host environment
• Wire tapping
– Insert a device into line or listen to wireless
– Passive (listen) or Active (modify)
• Waste searching
– Look for passwords or sensitive data
Operating Systems
8
Types of Attacks
• Masquerading
– Impersonate process, user, service
– Used from outside:
• Use stolen password (impersonate a legitimate user)
• Break communication line, assume session
– Used from within (spoofing):
• Impersonate login shell, steal password
• Trial and error
– e.g., try to guess password
Operating Systems
9
Types of Attacks -- Classification
• From within
– direct access as a valid process
– indirect Access via agent (attacker not present during attack)
• From outside
– channels provided for legitimate purposes
– illegitimate channels
Operating Systems
10
Functions of a Protection System
• External safeguards
– guard physical access (locks, badges, cameras)
• Verification of user identity (User Authentication) 
• Communication safeguards
– protect public/vulnerable lines: cryptography
• Access control (Ch 13)
– can Subject perform function on Resource
• Information flow control (Ch 13)
– can S get information contained in R (indirectly)
Operating Systems
11
User Authentication
• Approaches:
– Knowledge of some information
• Password, dialog
– Possession of some artifact
• Machine-readable cards (ATM)
• Combine with knowledge (PIN)
– Biometrics: Physical characteristics of person
• Fingerprint
• Hand geometry
• Face geometry
• Retina or iris scan
• Voice print
• Signature dynamics
Operating Systems
12
Fingerprint Recognition
• Extremely useful biometrics
technology
• Fingerprints are a primary and
accurate identification method
Operating Systems
13
Fingerprint Recognition
 Uses the ridge endings and bifurcation's to plot points known as
minutiae
 The number and locations of the minutiae vary from finger to
finger and from person to person
Finger Image
Operating Systems
Finger Image
+ Minutiae
Minutiae
14
Face Recognition
• Uses an image or series of images
• Principle: analysis of the unique shape, pattern and
positioning of facial features
• Passive: does not require a person’s cooperation
• Highly complex technology
Common approach:
Face geometry
Operating Systems
15
Voice Recognition
•
•
•
•
Not speech recognition, it is speaker recognition
Low-cost (cheap hardware)
Not very accurate (voice varies, noise)
Can be stolen (recorded)
Operating Systems
16
Hand geometry
• one of the most deployed biometrics world wide
Ben Gurion Airport
(Israel)
Operating Systems
17
Signature Verification
• Static/off-line:
– match pattern (image)
– can easily be reproduced
• Dynamic/On-line:
– match movement of the pen during signing process
(pressure, speed)
• Many commercial products
Operating Systems
18
Iris recognition
• Based on visible features, i.e. rings, furrows, freckles and the
corona
• Safest, most accurate biometrics technology
Heathrow Airport
(London)
Operating Systems
19
Retina recognition
• Capture the pattern of blood vessels throughout the
retina
• No two retinas are the same, even in identical twins
• More difficult/less convenient than iris scan
Operating Systems
20
Other techniques
• DNA
– Unique (except for identical twins) but
many imitations:
– not fully automated, slow, expensive
– privacy issue – DNA contains information
about race, paternity, medical conditions
– requires a physical sample of tissue
• Thermograms
– infrared camera to detect the heat patterns
Operating Systems
21
User Authentication
• Problem with biometrics:
uncertainty in recognition
– System generates a
number 0  n  1
– Bimodal distribution
– Threshold must be
chosen to minimize
• false alarms
• imposter acceptance
Operating Systems
22
User Authentication
• Passwords
– Must protect stored password files from access
– Must prevent trial and error (guessing)
• Protecting password files
1. Maintain unencrypted; rely on access control
2. Encrypt using “one-way” function H:
• H-1 is unknown
• knowing H(x) does not yield x
• keep only H(pw) with user name
• at login, compute H(pw’) and compare with H(pw)
Operating Systems
23
User Authentication
• Preventing password guessing
– System-generated pw
• Random string: difficult to memorize
• “Pronounceable” words
– System-validated
• Accept only passwords that obey specifications
(length, mix of letters/digits, upper/lower case)
• Employ password-cracking programs to
reject easy-to-guess passwords
– Time-limited
• Expiration date or number of uses
Operating Systems
24
User Authentication
• One-time passwords
1. Smart card (can be lost or stolen)
2. Use secret function;
• System generates a challenge n, user replies with
f(n) as password; e.g. f(n)=3*n/2
3. Use one-way function to generate series of
one-time passwords from one password pw
• … H(H(H(pw))) H(H(pw)) H(pw) pw
• Intruder can derive H(H(pw)) from H(pw)
but not H(pw) from H(H(pw))
because H-1 is unknown
Operating Systems
25
User Authentication
• guess any valid password:
– repeatedly generate strings s
(dictionary, random, …),
check if H(s) is in table
• System-extended pw
– for each pw, generate random
number slt (called “salt”)
– store: UserName,slt,H(slt,pw)
– guessing: must check H(slt,s)
for every slt in table
– salting does not make it
harder to guess the password
of a specific user
Operating Systems
26
Functions of a Protection System
• External safeguards
– Guard physical access (locks, badges, cameras)
• Verification of user identity (User Authentication)
• Communication safeguards

– Protect public/vulnerable lines: cryptography
• Access control
– Can Subject perform function on Resource
• Information flow control
– Can S get information contained in R (indirectly)
Operating Systems
27
Secure Communication
• Principles of cryptography:
– Cipher text, Plain text, Key(s)
– Encrypt: C = E(P,Ke)
– Decrypt: D(C,Kd) = P
• Goals:
– Secrecy = message content not revealed
– Integrity = message not modified
– Authenticity = establish identity of sender
– Nonrepudiability = establish identity of creator
(regardless of who sent it)
• an actor cannot deny creation of message (signature)
Operating Systems
28
Secure Communication
•
Secret-key Cryptosystems
Symmetric:
S and R share a common secret key K
which is used for both encrypting and decrypting
Operating Systems
29
Secure Communication
• Examples
• transposition cipher:
• rearranges the order of letters
• example algorithm: swap 2 letters, skip n
• key: n
• e.g., n=1: 'hello world' → 'ehlol owrdlnd’
• substitution cipher
• replace letters or groups of letters
• example: Cesar cypher
• algorithm: replace every letter by the letter k
positions down in the alphabet
• key: k
• e.g., k=1: 'fly at once' → 'gmz bu podf‘
• Easy to break using statistical analysis
Operating Systems
30
Secure Communication
• Example: DES
– US standard (1977)
– Blocks of 64 bits
– Block is divided into L and R half
– F applies Key to R
– result is XOR’d with L, becomes new R
– old R becomes new L
– repeat 16 times
– F uses:
• permutations
• substitutions
• XOR with a 56-bit key
• Triple-DES
C = DEA(DEA(DEA(P, K3), K2, K1)
Operating Systems
31
Secure Communication
• With Secret-key cryptosystems:
– Secrecy: only R can decrypt C
– Integrity: intruder cannot produce valid message
– Nonrepudiation: not possible, S can deny
– Authenticity of sender: must prevent replay
Operating Systems
32
Secure Communication
• Use nonce N to prevent replay of message:
S
R
(1)
N
(2)
C=E({P,N},K) 
– Capturing either message does not help;
both are different every time
• Use timestamp T to prevent replay
S
C=E({P,T},K) 
R
– Limits possible replay to a chosen time interval
Operating Systems
33
Secure Communication
• Key distribution and authentication
– Both S and R must have the same key K
– Trusted server approach:
• Each process has its own secret key for communication
with trusted Key Distribution Center (KDC)
• At runtime, process A asks KDC for a Session Key
KAB for communication with process B
KDC
(1)
 A,B
(2)
E({KAB,B,ticket},KA) 
(3)
ticket = E({KAB,A},KB)
Operating Systems
A
B
ticket 
34
Secure Communication
• Public-key cryptosystems (Diffie-Hellman, 1976)
– Asymmetric: different keys for encryption and
decryption
– One cannot be derived from the other
– One is Public key, other is Private
Operating Systems
35
Secure Communication
• With Public-key cryptosystems
– Secrecy: only R can decrypt message using KRpriv
– Integrity: intruder cannot produce valid message without
KSpriv
– Authenticity of creator: same as integrity: only S knows
KSpriv
– Authenticity of sender: use nonce or timestamp to prevent
replay
Operating Systems
36
Secure Communication
• How to make a key/function so that the other cannot be
derived from it?
• RSA (Rivest, Shamir, Adelman) Public Key
C = E(P) = Pe mod n
P = D(C) = Cd mod n
– (e,n): Public encryption key
– (d,n): Private/secret decryption key;
d cannot be derived from e
Operating Systems
37
Secure Communication
• RSA Key Generation
1. Choose large primes p and q; compute n=p*q
• Example: p=5, q=7, n=35
2. Choose d to be a (large) prime number having
no factors in common with (p1)*(q1)
• Example:
(51)*(71)=24; d=5 or 7 or 11 (choose 11)
3. Choose e such that e*d mod (p1)*(q1) = 1
• Example:
e*11 mod 24 = 1; e = 11 or 35 or 59 or 83 …
C = E(P) = P59 mod 35
P = D(C) = C11 mod 35
Operating Systems
38
Secure Communication
• Why is RSA encryption secure?
– n is derived from p and q; (n=p*q)
– d is also derived from p and q; (no common factors)
– e is derived from d but also needs p and q;
– only d is known/public, p and q have been discarded → e
cannot be derived
– similarly, d cannot be derived from e without p and q
Operating Systems
39
Secure Communication
• Public key distribution and authentication
– Making key public is easy, but need to authenticate it:
• How does A safely get B’s public key KBpubl ?
• Trusted server approach:
KDC
(1)
(2)
 A,B
E({B,KBpubl},KKDCpriv) 
A
– KDC provides B’s public key KBpubl
– KKDCpriv guarantees authenticity (KDC sent it)
Operating Systems
40
Secure Communication
• Digital Signatures:
– How can a document be “signed” and transmitted
electronically?
– Here is my signature
• Anyone can copy and attach it to any document
– Sign on paper, scan
• Any document is digitized and can be modified
– Public-key cryptosystems permit unforgeable electronic
“signatures”?
Operating Systems
41
Secure Communication
• Digital Signature: document M is to be “signed”
– Sender generates unique digest: d = H(M)
– Sender encrypts E(d,KSpriv), receiver decrypts with KSpubl
– Receiver computes d’ = H(M);
• d’ is a unique signature of document M
• d=d’ means that d is a also a unique signature of M;
• Decryption authenticates sender, proving sender sent d
i.e., sender signed M
Operating Systems
42