Summary From the Last Lecture

1. Summary From the Last Lecture • Security is multifaceted phenomenon • Confidentiality, integrity, availability • We spoke about various security threats • And some general defense approaches • Prevention • Detection and response (sustain the attack or get rid of it) • Learn from mistakes, improve prevention • Now we’ll talk about challenges in the defense field

2. What Are the Challenges? • Your security frequently depends on others • Tragedy of commons • A good solution must • Handle the problem to a great extent • Handle future variations of the problem, too • Be inexpensive • Have economic incentive • Require a few deployment points • Require non-specific deployment points

3. What Are the Challenges? • Fighting a live enemy • Security is an adversarial field • No problem is likely to be completely solved • New advances lead to improvement of attack techniques • Researchers must play a double game

4. What Are the Challenges? • Attack patterns change • Often there is scarce attack data • Testing security systems requires reproducing or simulating legitimate and traffic • No agreement about realistic traffic patterns • No agreement about metrics • There is no standardized evaluation procedure • Some security problems require a lot of resources to be reproduced realistically

5. Practical Considerations • Risk analysis and risk management • How important it is to enforce a policy • Which threats matter • Legislation may play a role • The role of trust • Assumptions are necessary • Human factors • The weakest link

6. In The Shoes of an Attacker • Motivation • Bragging Rights • Profit (Spam, Scam, Phishing, Extortion) • Revenge / to inflict damage • Terrorism, politics • Risk to the attacker • Usually small • Can play a defensive role

7. Why We Aren’t Secure • Buggy code • Protocol design failures • Weak crypto • Social engineering/human factor • Insider threats • Poor configuration • Incorrect policy specification • Stolen keys or identities • Misplaced incentives (DoS, spoofing, tragedy of commons)

8. The Role Of Policy • Policy defines what is allowed and how the system and security mechanisms should act • Policy is enforced by mechanism which interprets and enforces it, e.g. • Firewalls • IDS • Access control lists • Implemented as • Software (which must be implemented correctly and without vulnerabilities)

9. Some Security Mechanisms • VPNs • Intrusion Detection • Intrusion Response • Development tools • Virus scanners • Policy managers • Trusted hw • Encryption • Checksums • Key management • Authentication • Authorization • Accounting • Firewalls

10. Cryptography

11. What Is Cryptography? • Goal: Protect private communication in the public world • Alice and Bob are shouting messages over a crowded room • Everyone can hear what they are saying but no one can understand (except them) • We have to scramble the messages so they look like nonsense or alternatively like innocent text • Only Alice and Bob know how to get the real messages out of the scramble

12. Cryptography Is Also Useful For … • Authentication • Bob should be able to verify that Alice has created the message • Integrity checking • Bob should be able to verify that message has not been modified • Non-repudiation • Alice cannot deny that she indeed sent the message

13. Cryptography Is Also Useful For … • Exchanging a secret with someone you have never met, shouting in a room full of people • Proving to someone you know some secret without giving it away • Sending secret messages to any m out of n people so only those m can retrieve messages and the rest n-m cannot • Sending a secret message so that it can be retrieved only if m out of npeople agree to retrieve it

14. So, How Do We Scramble Messages? • Alice could give a message covertly “Meeting at the old place” • Doesn’t work for arbitrary messages and • Doesn’t work if Alice and Bob don’t know each other • Alice could hide her message in some other text – steganography • Alice could change the message in a secret way • Bob has to learn a new algorithm • Secret algorithms can be broken by bad guys Good cryptography assumes knowledge of algorithm by anyone, secret lies in a key!!!

15. Sample Crypto Scheme: Ceasar’s Cipher • Substitute each letter with a letter which is 3 letters later in the alphabet • HELLO becomes KHOOR • Instead of using number 3 we could use n[1,25]. nwould be our key • How can we break this cipher? Can you decipher this: Bpqakzgxbwozixpgammuazmitmiag. Emeivbuwzm!

16. Sample Crypto Scheme: Ceasar’s Cipher • We can also choose a mapping for each letter:(H is A, E is M, L is K, O is Y). This mapping would be our key. This is monoalphabeticcipher. • HELLO becomes AMKKY • How can we break this cipher?

17. Types Of Cryptographic Functions • Symmetric key crypto: one key • We will call this secret key or shared key • Both Alice and Bob know the same key • Asymmetric key crypto: two keys • Alice has public key and private key • Everyone knows Alice’s public key but only Alice knows her private key • One can encrypt with public key and decrypt with private key or vice versa • Hash functions: no key

18. Uses Of Symmetric Key Crypto • Symmetric key crypto: one key • Transmitting over an insecure channel • Classic use: Alice and Bob encrypt messages they exchange • Secure storage on insecure media • Encrypt stored data so someone who breaks in cannot read it

19. Uses Of Symmetric Key Crypto • Authentication – prove the identity • Pass phrase – what if Mallory asks for the pass phrase • Strong authentication without revealing the secret • Integrity check • Calculate the checksum and encrypt it – MIC: message integrity code Bob Alice RA KAB(RA) RB KAB(RB) M MIC = EAB(check(M))

20. Uses Of Asymmetric Key Crypto • Asymmetric key crypto can do everything symmetric key crypto can but much (about 1,500 times) slower • However, it can do some things better! • However, it can do some extra things! • Transmitting over an insecure channel • If Alice wants to talk to Bob and Carol using symmetric key crypto she either has to remember two keys or run a risk that Bob can impersonate her when talking to Carol • With asymmetric keys, Bob and Carol know Alice’s public key and Alice knows her private key

21. Uses Of Asymmetric Key Crypto • Secure storage on insecure media • Same as with symmetric key crypto • Authentication • Alice wants to verify Bob’s identity • She sends to Bob EPubBob(RA) • Bob decrypts and sends back RA • This can be done with symmetric keys too but if Bob wanted to authenticate himself to Carol he would need to remember a new key. Not so with asymmetric keys. • Alice doesn’t need to store any secret info which is good if she is a computer

22. Uses Of Asymmetric Key Crypto • Digital signatures • Alice orders books online from Bob • She signs every order using her private key • If she claims she didn’t place the order Bob can prove she did – non-repudiation • Can symmetric key crypto do this?

23. Hash Algorithms • Known also as one-way functions or message digests • Take an arbitrary-length message M and transform it into fixed-length hash h(M) • Properties: • Knowing M is easy to calculate h(M), but it is very hard to calculate M knowing h(M) • It is very hard to find M1 M so that h(M1) =h(M), this is collision-free property • E.g., take the message M as a number, add a large constant to it, square it, and take middle ndigits as the hash

24. Uses Of Hash Algorithms • Storing hashed password info • Message integrity • Use message M and a shared secret S, run this through hash function and produce MIC • Send only M and MIC • Why do we need a shared secret? • Message fingerprint • Hash the files to detect tampering • Works for download security too • Signing message hash instead of the whole message is faster

25. Crypto TerminologyCryptographic TechniquesStream Ciphers

26. Let’s Formalize A Little … M C M Bob Alice EK1(M) DK2(C) K1 K2 M – message K1 – encryption key EK1(M) – message M is encrypted using key K1 C – ciphertext K2 – decryption key DK2(C) – ciphertext C is decrypted using key K2 If K1=K2 this is symmetric (secret key) encryption If K1K2 this is asymmetric (public key) encryption

27. Attacks M C M Bob Alice EK1(M) DK2(C) K1 K2 Eve Cyphertext-only attack: Eve can gather and analyze C’s to learn K2 How does Eve know she got the right key?Eve has to have enough ciphertext – having XYZ withmonoalphabeticcipher would not be enoughWhat if K2 depends on a password in a known way?

28. Attacks M C M Bob Alice EK1(M) DK2(C) K1 K2 Eve Known-plaintext attack: Eve can attempt to learn K2 by observing manyciphertextsC for known messages M How does Eve obtain the plaintext?

29. Attacks M C M Bob Alice EK1(M) DK2(C) K1 K2 Mallory Chosen-plaintext attack: Mallory can feed chosen messages M into encryption algorithm and look at resulting ciphertexts C. Learn either K2 or messages M that produce C. Assumption is that extremely few messages M can produce same C. For a monoalphabetic cipher she could feed a message containing all the letters of the alphabet What if Alice has a limited vocabulary?

30. Attacks M C M Bob Alice EK1(M) DK2(C) K1 K2 Mallory • Man-in-the-middle attack: • Mallory can substitute messages • Mallory can modify messages • So that they have different meaning • So that they are scrambled • Mallory can drop messages • Mallory can replay messages to Alice, Bob or the third party

31. Attacks M C M Bob Alice EK1(M) DK2(C) K1 K2 Eve Brute-force attack: Eve has caught a ciphertext and will try every possible key to try to decrypt it. This can be made infinitely hard by choosing a large keyspace.

32. Cryptographic Techniques • Substitution • Goal: obscure relationship between plaintext and ciphertext • Substitute parts of plaintext with parts of ciphertext • Transposition (shuffling) • Goal: dissipate redundancy of the plaintext by spreading it over ciphertext • This way changing one bit of plaintext affects many bits of the ciphertext (if we have rounds of encryption)

33. Substitution • Monoalphabetic – each character is replaced with another character • Ceasar’s cipher – each letter is shifted by 3, a becomes d, b becomes e , etc. • Keep a mapping of symbols into other symbols • Drawback: frequency of symbols stays the same and can be used to break the cipher

34. Substitution • Homophonic– each character is replaced with a character chosen randomly from a subset • Ciphertext alphabet must be larger than plaintext alphabet – we could replace letters by two-digit numbers • Number of symbols in the subset depend on frequency of the given letter in the plaintext • The resulting ciphertext has all alphabet symbols appearing with the same frequency

35. Substitution • Polygram– each sequence of characters of length n is replaced with another sequence of characters of length n • Like monoalphabetic cipher but works on n-grams