where are we at topic overview
play

Where are we at - Topic overview Lecture 1A: Security - PowerPoint PPT Presentation

Where are we at - Topic overview Lecture 1A: Security requirements/features Lecture 7A Threatens Privacy Threatens Try to achieve Lecture 2B: Network threats Lecture 3A: Attacks on Lecture 6A: Security Protocols Web servers, malware


  1. Where are we at - Topic overview Lecture 1A: Security requirements/features Lecture 7A Threatens Privacy Threatens Try to achieve Lecture 2B: Network threats Lecture 3A: Attacks on Lecture 6A: Security Protocols Web servers, malware Use Lecture 5A&B Authentication Lecture 1B,2A: Cryptography 3B: Certificates and Trust Lecture 4A&B Access Rights 1

  2. Certificates & Trust Hashes, Digital Signature, PKIs, Trust management

  3. Error correction - Hash - MAC  Excerpt; short `description’ of document  Fixed size output for any size input 134396e4399b7e753ffca7ba366c418f gimp-2.8.0-RC1.tar.bz2 28997d14055f15db063eb92e1c8a7ebb gimp-2.8.0.tar.bz2  CRC check or MD5 checksum  Common for e.g. ftp sites  Does this add security? 3

  4. Error correction - Hash - MAC  Excerpt; short `description’ of document  Fixed size output for any size input  Goals  Integrity: message not altered  Authentication: message from X  Proof of possession without revealing content now  Non-repudiation 134396e4399b7e753ffca7ba366c418f gimp-2.8.0-RC1.tar.bz2 28997d14055f15db063eb92e1c8a7ebb gimp-2.8.0.tar.bz2 4

  5. Properties of Hash functions 1-Way `random function’ Pre-image resistant Collision resistant Second pre-image resistant m m H(m) H(m) = = m H(m) H(m’) H(m’) m’ m’ Hard to find: m with H(m) = h m, m’ with H(m) = H(m’) m’ with H(m’) = H(m) Practical Efficiently computable m H(m) 5

  6. Applications of Hash functions  Message Digest  Check have correct message  Password storage  No reverse; how verified?  Password recovery?  Message Signing  Signing large message is slow  Sign hash of message instead 6

  7. An Example: MD5 Message padded so total size is multiple of 512 bits 64 bits message padding: 10..00 length 512 bits 512 bits 512 bits Y output IV CVi CVi+1 HMD5 HMD5 HMD5 128 128 128 128 128 128 Compression Function 7

  8. Compression functionHMD5 CV(in) Y (block) 128 512 X P1 RF(F) Round Permutations Permutations Permutations Permutations P2 RF(G) Function (next slide) P3 RF(H) P4 RF(I) 128 CV(out) 8

  9. RF(F) Modular addition  For k=1 to 16 do A B C D Permuted Text Block F Chaining Chaining Value (CV) Value (CV) X[k] T[k] S[k] Array of Array of A B C D Constants Constants 9

  10. Weakness MD5  http://www.win.tue.nl/hashclash/rogue-ca/  video at e.g. http://dewy.fem.tu-ilmenau.de/CCC/25C3/ video_h264_720x576/25c3-3023-en- making_the_theoretical_possible.mp4 10

  11. Message Authentication Codes  Unable to predict for unseen message  Keyed; validation requires same key  Authenticity and Integrity  Example:  Keyed-Hash; uses (symmetric) key  Hmac; masked key pre-pended before hash. Key Generation & Message MAC Validation (any length) 11

  12. Digital Signatures  `Public key version of a MAC’  Signing with a private key  Decryption of Hash of Message  Verification with public key Message Message (any length) (any length) Private Key Public Key ? Decrypt Digital = Hash Encrypt Hash (=Sign) Signature Generation Validation 12

  13. Digital signatures with RSA Public key Private key Alice: KA Alice: kA Signing Message M: Checking Signature: Compute hash h := H(M) Compute hash h := H(M) Signature s := RSA_D(kA, h) Check: RSA_E(KA, s) == h Uses fact: RSA_E( KA, RSA_D(kA, x) ) = x 13

  14. Key distribution Private key Public key Signature Alice Bob 14

  15. (Wo)man-in-the-middle attack Private key Public key Eve Signature Alice Bob 15

  16. Bob’s public key is 1234 Certificate Bob is a Baker ... EXP DATE: 29-2-2013 Statement (e.g. Identity, Attribute) signed by principal whom believes it to be true at time of signing and/or: assumes responsibility, liability, … Example: X.509 - Statement links a key to attributes Note: Revocation; Validity period – revocation certificate 16

  17. Trust me I’m a doctor Certificate based Trust Management Trust based on formal relationships PoFI 2010 Feb 5 th 2010

  18. Certificate Authority CA E.g. Verisign Verification method? Root CA validate, certify Intermediate CA Intermediate CA Intermediate CA validate, Validates attribute certify Identity, role, e-mail address, Intermediate CA Intermediate CA Web address, etc. validate, Links them to pub key certify pub key - pub key - pub key - attributes attributes attributes (Demo Certificates and CA in browser) 18

  19. Transitive and full trust  Dec12/Jan 13: Turktrust fake certificate discovery  Fake intermediate CA certificates (issued august 11)  Aug11: Hack DigiNotar confirmed  Dutch Certificate Authority  First hack already in June 2011  Many rogue SSL certificates  (Diginotar bankrupt in September 2011)  March11: Comodo partner incidient  9 fake certificates issued (e.g. live/google/yahoo/skype/mozilla)  quickly discovered and disseminated. CA can Issue any certificate. 19

  20. Web of Trust  Recall First Lab session  Validate key directly  New keys signed by known keys  No centralized CAs  Each user signs keys they trust  User can choose degree of trust in other keys  For communication  For signing other keys Compare S/MIME – CA signed certificates 20

  21. Rule based trust management  Generalizes tree of CAs Policy rules Alice: Meaning: A.r ← B Alice trusts Bob in role r (Bob is certified for r) A.r ← B.r Alice trust Bob certifying r (Bob is a CA for r) A.r ← A.cert.r Alice trusts anyone in A.cert to certify r (Everybody in A.cert is CA for r) A.r ← B.r /\ C.r Alice trusts if both Bob and Charlie trust. Can also use multiple different roles `r’. 21

  22. Certificate, Rule Based Trust  Policy: GMS.Dr may read Patient record Green  Rules to establish Doctors Medical Service GMS.Dr ← GMS.Department.Dr GMS.Department ← Radiology Radiology.Dr ← Alice  Alice may read the patient record  Trusted, Certified facts & Delegation 22

  23. 1.000 satisfied customers and counting. You can trust me, ask anyone Reputation based Trust Management Trust based on opinions/experiences

  24. Fast Reputation, Behaviour Based Trust and Quite Good  Policy Dr with good reputation may treat  Reputation based on Past Performance  Feedback after interaction updates reputation  E-bay, Eigentrust, pagerank, centrality measures  Estimate risk based on Reputation  Good reputation valuable  Incentive for good behaviour 24

  25. Combined Trust Scores Certificates Fast and Quite Good Other sources… Feedback & Recommendations User’s Requirements determine how to mix TRUST 100% 25

Recommend


More recommend