IS 302: Information Security and Trust Week 9: User Authentication (part II) and Introduction to Internet Security 2012
© Yingjiu Li Review Review of weak pwd authentication Unix pwd Windows password (LM and NT hashes) Attacks (dictionary and brute-force)
© Yingjiu Li Strong Password Authentication I.Lamport scheme II.Challenge response III.Time stamp To address eavesdropping and replay attacks in communications
© Yingjiu Li I. Lamport Scheme Registration phase –H: one way hash function –Bob: choose a secret s; compute and send w0=H^100(s) to Alice in a secure manner Bob Alice Mallory s H^100(s) Bob, w0=H^100(s)
© Yingjiu Li Lamport scheme: first-time authentication –Bob sends one-time password w1=H^99(s) to Alice –Alice authenticates Bob by checking if H(w1)=w0 –If match, Alice replaces w0 with w1=H^99(s) Bob Alice Mallory s w1=H^99(s) Bob, w0=H^100(s)
© Yingjiu Li Lamport scheme: second time authentication –Bob sends one-time password w2=H^98(s) to Alice Alice authenticates Bob by checking if H(w2)=w1 If match, Alice replaces w1 with w2=H^98(s) Bob Alice Mallory s w2=H^98(s) Bob, w1=H^99(s)
© Yingjiu Li Discussion Why is Lamport scheme secure against eavesdropping and replay attacks? How many times can Alice authenticate Bob in Lamport scheme? After that?
© Yingjiu Li II. Challenge Response Scheme Alice and Bob share a pwd or pwd hash S Upon receiving Bob’s id and authentication request, Alice sends Bob a non-repeating challenge C Bob responses with R=E(S,C) or H(S,C) Bob Alice Mallory Bob id SS Non-repeating challenge C Response R=E(S,C) or H(S,C)
© Yingjiu Li Discussion Why is challenge response scheme secure against eavesdropping and replay attacks? What is the difference between challenge response scheme and Lamport’s scheme?
© Yingjiu Li Case Study: MS Authentication Protocol (NTLM) Windows 2000, NT, XP, Vista, Windows 7 clientserver Type 1 msg: a set of flags (enc key sizes, req for mutual auth) clientserver Type 2 msg: a set of flags + random challenge (8 bytes) clientserver Type 3 msg: server response = MD4/MD5 hashing or DES enc (server challenge + client pwd hash)
© Yingjiu Li NTLMv1 and NTLMv2 LM-hash of pw: 16 bytes (DES+DES) NT-hash of pw: 16 bytes (MD4) C= 8-byte server challenge, random K1 | K2 | K3 = NT-hash | 5 bytes-0 R1 = DES(K1,C) | DES(K2,C) | DES(K3,C) K1 | K2 | K3 = LM-hash | 5 bytes-0 R2 = DES(K1,C) | DES(K2,C) | DES(K3,C) Response = R1 | R2 CS= 8-byte server challenge, random CC=8-byte client challenge, random CC* =8-byte client information (X, time, CC, domain name) v2-Hash = HMAC-MD5(NT-hash, user name, domain name) LMv2 = HMAC-MD5(v2-hash, CS, CC) NTv2=HMAC-MD5(v2-hash, CS, CC*) Response = LMv2 | CC | NTv2 | CC*
© Yingjiu Li III. Time Stamp Scheme Alice and Bob keep synchronized clocks –Time T is used as one-time “challenge” –Bob sends Alice his id, and R=E(S,T) or H(S,T) –Alice verifies R=? E(S,T) or H(S,T) on Bob’s pwd (or pwd hash) S and her current time T Bob Alice Mallory Bob id, R=E(S,T) or H(S,T) SS
© Yingjiu Li Discussion Why is time stamp secure against eavesdropping and replay attacks? How do you compare time stamp scheme with challenge response scheme and Lamport scheme?
© Yingjiu Li Break
© Yingjiu Li Introduction to Internet Security I.Secure Socket Layer (SSL) II.Firewall (FW) III.Intrusion Detection System (IDS)
© Yingjiu Li I. SSL Secure socket layer (SSL) or transport layer security (TLS) Secure communications between clients and server in internet TCP SSL web applications Application protocol SMTP,HTTP,FTP… TCP IP Network Interface TCP/IP Protocol Stack SSL
© Yingjiu Li SSL functionalities –Server authentication with certificate –Optional client authentication –Encrypted channel between server and client
© Yingjiu Li SSL HandShake Protocol Web client Web server 1.SSL Request 2.Server public key certificate 3. random number R encrypted using server public key 4. Both compute session key based on R and switch to encrypted tunnel (https)
© Yingjiu Li Typical Applications with SSL E-commerce VPN
© Yingjiu Li II. FW Filter and audit traffic between internal and external network Prevent unauthorized access to a private network Internal network External network Firewall
© Yingjiu Li FW Policies Close policy: Default deny, allow (white list) firewall (p: packet) { if (allow(p)) forward(p); else drop(p); } Open policy: Default permit, deny (black list) firewall (p: packet) { if (deny(p)) drop(p); else forward(p); }
© Yingjiu Li Close Policy Example boolean allow(packet p){ if (match(p.srcIP, *.*) and match (p.dstPort, 25)) return true; elseif (match(p.srcIP, *.*) and match (p.dstPort, 80)) return true; else return false; }
© Yingjiu Li Open Policy Example boolean deny(packet p){ if (match(p.srcIP, *.*) and match (p.dstPort, 25)) return true; elseif (match(p.srcIP, *.*) and match (p.dstPort, 80)) return true; else return false; }
© Yingjiu Li Windows FW –start cmd firewall.cpl advanced settings Inbound rules Outbound rules
© Yingjiu Li Limitation of FW Internal network External network Firewall
© Yingjiu Li III. Intrusion Detection System (IDS) IDS should alert an attack in progress –With high accuracy –In real time manner –With complete diagnosis –With effective recommendations on how to react
© Yingjiu Li Accuracy of IDS False positive (alarm) rate –#false-alerts/#alerts False negative (miss) rate –#missed-attacks/#attacks
© Yingjiu Li Signature-Based Vs Anomaly-Based Signature based IDS –Signature: DB of known attack patterns –IDS reports situations that match signatures –Good: Low false alarm rates, instantaneous detection –Bad: Cannot detect new attacks Anomaly based IDS –Profile: model of normal behaviors –IDS reports situations that deviate from profiles –Good: Can detect some new attacks –Bad: High false alarm rates, high complexity
© Yingjiu Li 2007INFT 865 Ravi Sandhu Why false alarm rate is high in anomaly-based detection Analogy –Test for a disease is 99% accurate 100 disease-free people tested, 99 test negative 100 diseased people tested, 99 test positive –Prevalence of disease is 1 in 10,000 –Alice tests positive –What is probability Alice has the disease? 1 in 101 (101=1*99%+9,999*1%) –False alarm rate: 100 in % Now assume that test is 99.98% accurate –What is the false alarm rate?
© Yingjiu Li IDS Summary Advantages –Complement firewall –Continues to improve Disadvantages –False alarms –Miss some new attacks –Require security personnel to handle alarms and monitor track records
© Yingjiu Li Review How long is NTLM (v1) response message? 1.64 bits bits bits Which of the following must have a cert in SSL? 1.Client 2.Server 3.Both To detect new attacks, you need to choose 1.Firewall 2.Signature-based IDS 3.Anomaly-based IDS