NETWORK SECURITY... Authors Hari Thiruvengada Varun Lalchandani Contents • • • • • • • • • • Introduction. Encryption and Protocols. Kerberos. Network Security Issues and Protocols. Intrusion Detection Web Security. Digital Identification Techniques. Cryptography and web Security. Firewalls. References. Security and its breaches… Security • a system is secure if it is – Security goals are achieved. – Components behaves as expected on it. Breaches • • • • Interruption - System asset lost , unavailable or unusable. Interception - Unauthorized party gains access to asset. Modification - Tampering with the asset. Fabrication - counterfeit objects on computing system. Security Goals and Vulnerabilities Security Goals • Confidentiality - assets of a computing system accessible only by authorized user.Read only type of access like viewing, printing helps in privacy. • Integrity - modification only by authorized parties. Precise, accurate, consistent assets. • Availability - assets are accessible to authorized parties.Timely response, fair allocation, fault tolerance, usability, controlled concurrency. (Denial of service attacks.) The people involved… • Amateurs - fresh players of the game , disgruntled over a -ve work situation. • Crackers - breaking into unauthorized territory without malicious intent. • Hackers - breaking into unauthorized territory with malicious intent. • Career Criminals - people in the game for money and have predefined targets. Basic Encryption and Decryption • Encryption - process of encoding a message so that its meaning is not obvious. • Decryption - process of decoding the encrypted message. • Cryptography - Hidden writing, which conceals meaningful text. • Cryptanalyst - studies encryption and finds hidden messages. • Cryptanalysis – attempt to break a single message. – Recognize patterns in encrypted messages to break into subproblems by straightforward decryption algorithm. – Find weakness in encryption algorithms. Contd… Encryption Plain Text Original Text Cipher Text Encryption Encryption with Keys Decryption (Symmetric Cryptosystem) Key Plain Text Original Text Cipher Text Encryption Encryption with Keys Encryption Key (Ke) Plain Text Decryption (Asymmetric Cryptosystem) Decryption Key (Kd) Original Text Cipher Text Encryption Decryption Encryption • Substitutions - one letter x-changed for other. – Monoalphabetic Ciphers. • Caesar Cipher Example: Plaintext:ABCDEFGHIJKLMNOPQRSTUVWXYZ Cipher :DEFGHIJKLMNOPQRSTUVWXYZABC – Polyalphabetic Ciphers. • Frequency distribution reflects the underlying letters. Table for Odd Positions ABCDEFGHIJKLMNOPQRSTUVWXYZ ADGJMPSVYBEHKNQTWZCFILORUX Table for Even Positions ABCDEFGHIJKLMNOPQRSTUVWXYZ N S X C H M R WB G LQ VAF K PU Z E J O TYD I Contd… Example: F1(x) = (3 * x)mod 26. F2(x) = ((5 * x) +13) mod 26. Encryption for : TREAT YIMPO SSIBL E would be FUMNF DYVTF CZYSH H Contd… • Transposition - letters of message rearranged. – GOAL - Diffusion Example: C1 C2 C6 C7 C11 C12 C3 C8 C4 C9 C5 C10 C4 C9 C5 C10 etc. The resulting cipher text will be C1 C6 C11 C2 C7 C12 C3 C8 etc. Encryption Protocols… Arbitrated Protocol B Arbiter A A acted fairly Adjudicated Protocol A B Adjudicator You are cheating Self-Enforcing Protocol A B Symmetric Key X-change W/out Server • Small no. of messages. • Less risk of intrusion. • Each user have a copy of symmetric secret encryption key K. • For greater security one can generate a fresh key called KNEW. • Send E(KNEW,K). Symmetric Key X-change With Server DISTRIBUTION CENTER(DC) 1.Give me a key 2. E((KPR ,P), KR ) (P, R, Ip ) 2.Here is the key and E(Ip ,R,KPR,,E((KPR ,P), KR)), KP) Renee 3. DC gave me key for Private Communication E((KPR ,P), KR ) Pablo Asymmetric Key X-change W/out Server 1. EP(DR (K)) Renee sends new key 2. S (n,K) Pablo sends encrypted random number Renee 3. S (n+1,K) Renee returns successor of Random Number Pablo Asymmetric Key X-change With Server DISTRIBUTION CENTER(DC) 4. Please give me Pablo’s public key (R ,P) 5. Here’s Pablo’s Key (DD(EP ,P)) 1.Give me a Renee’s key (P,R) 2.Here is the Renee’s Key (DD(ER,R)) 3. I’ am Pablo, let’s Talk. ER(P, IP) 6. Renee here , What’s up ? EP(R, IR ) 7. Reply with ER(M, IR) Renee Pablo Authentication in Distributed Systems • Kerberos Initiating a Kerberos Session User U 1. U’s Identity 2. Encrypted under Password Session key SG Ticket TG Kerberos Server Ticket Granting Server 2. Encrypted under KS-TGS Key Session key SG Contd.. Obtaining a ticket to access a file 1. Request to Access File F User U Ticket Granting Server Ticket to File Server to access File F + SF 2.Encrypted under TGS - F Key + SF How Kerberos withstand Attacks? • No passwords communicated on Network. • Cryptographic protection. • Limited period of validity. • Mutual authentication. Why Kerberos is not the perfect Answer? • Kerberos requires the availability of continuous trusted “Ticket Granting Server ”. • Trusted relationship required between TGS and every server. • Requires timely transactions. • Subverted workstation can save and later replay user passwords. • Password guessing works. • Does not scale well. NETWORK SECURITY ISSUES • • • • • • Sharing. Complexity of system. Unknown Perimeter. Many points of attack. Anonymity. Unknown path. Threats • • • • • • • Wiretapping. Impersonation. Message Confidentiality Violations. Message Integrity Violations. Hacking. Code Integrity Violation. Denial of Service. Wire tapping C Microwave link A Inductance Packet Sniffer B Network Security Protocols • Link Encryption. sender Receiver Intermediate Router Protocol Layers Message encrypted Message in plain text. Contd... • End to End Encryption. sender Receiver Intermediate Router Protocol Layers Message encrypted Message in plain text. Comparisons Link Encryption End to End Encryption 1.Message X-posed in sending host / intermediate routers. 2.Applied by sending host. 3.Invisible to user process. 4.Host maintains encryption. 5.can be done in H/w. 6.All / No message encrypted. 7.One key per host pair. 1.Message encrypted in sending host / intermediate routers. 2.Applied by sending process. 3.User selects algorithm. 4.User selects encryption. 5.S/w implementation. 6.User chooses to Encrypt / Not. 7.One key per user pair. Intrusion Detection • Intrusion - set of actions attempts to compromise integrity,confidentiality or availability of resources. • IDS based on Data source -host based : audit data from single host. -multi host based : audit data from multiple host. -network based : network traffic data along with audit data from one or many host. • IDS based on Model of Intrusion. -Misuse detection system: look for the exploitations of known weak points. -Anomaly Detection System: detect changes in the pattern of utilization or behavior of system. Desirable Characteristic of ID • • • • • Must run continually. Must be fault tolerant. Must resist subversion. Must impose minimal overhead. Must be able to adapt changes in the behavior of user and system. • Must be scalable. • Must provide graceful degradation. • Must allow dynamic reconfiguration. Limitations of Existing ID • Central Analyzer is single point of failure. • Scalability is limited. • Difficult to add capability or reconfigure. • Analysis of Network data can be flawed. Intrusion Types • • • • • • Attempted Break-in. Masquerade attack. Penetration of security control system. Leakage. Denial of service. Malicious use. ARCHITECTURE TO GENERATE THE STD. FORMAT LOG GENERATOR LOG FILTER LOG FILTER ANALYIS ENGINE LOG PROCESSOR LOG GENERATOR STANDARD AUDIT TRAIL FORMAT • Must satisfy two basic properties: Extensibility: Neither the names nor the number of fields of the log record are fixed. Portability : The log can be processed on any system. • SUMMARY OF THE STD. LOG FORMAT #S# start log record. #Fc# change field separator to c. #E# #N# # end log record. next log record. default field sep. #Cc# change nonprinting delimiter to c. #I# ignore next field. \ default nonprinting delimiter. The standard log contains fields. Each field is associated with an attribute. SunOS MLS Logs • A simplified example of a SunOS MLS log record is given as: header, 120, AUE_UNLINK, Wed Sep 18 11:35:28 1999, process, bishop, root, root, daemon, 1234, label, confidential, nuclear, crypto pathname, /, / usr / holly,…/ matt / tmp / junkfile return, Error 0, 5 trailer, 120 +57000ms • Put into the standard log format , this looks like: #S#event=AUE_UNLINK#date=09181999@113528#usedtime=57000#I# #logid=bishop#ruid=root#euid=root#rgid=daeon#procid=1234#I# #seclevel=confidential#class=nuclear#class=crypto#I# #rootdr=/#cwd=/usr/holly#pathname=../matt/tmp/junkfile#I# #errno=0#retval=5#E# Typical Anomaly Detection System Update profile statistically Audit Data Attack State System Profile deviant Generates new profile Anomaly ID (Statistical Approach) • Let S1, S2, S3 … Sn, represent abnormality values of profile measures M1, M2, M3 … Mn respectively. • Higher value of Si indicates greater abnormality. • A combining function of individual S values will be, a1S12 + a2S22 + a3S32 + … + anSn2 , a i > 0. Pros and Cons of Anomaly ID. • Statistical techniques have applicability here. • Statistical patterns could be used to capture patterns unique to the user. Types of Measure. • Activity Intensity Measure – measures the rate at which activity is progressing. E.g. no. of audit records processed per minute. • Audit record distribution measure – measures the all activity types in recent audit records.E.g. I/O activities. • Categorical Measure – measures the distribution of particular activity over categories. E.g. relative frequency of logins,relative usage of compiler, shells,editor etc. • Ordinal Measure – measure activity whose output is in numeric value. E.g. CPU Typical Misuse Detection System Modify existing rule Audit Data Rule System Profile match Add new rule Timing Info Attack state Misuse ID • Detection of intrusions by precisely defining them well ahead of time and watching for their occurrence. • Intrusion signatures are sequence of events and conditions that lead to a break-in. • Abstract high quality patterns from attack scenarios. • Should be simple enough to keep the matching tractable and should be compared. • If pattern matches then issue an alarm warning. Pros and Cons of Misuse ID • Looks only for known vulnerabilities, comparison is finite. • Little use to detect little known future intrusion patterns. Continues… Model based Intrusion Detection A Pattern Matching approach • Consider an initial set of 100 measures as potentially relevant to predicting intrusions. • Resultant is a set of 2100 possible measures. • Appropriate set of measures depend on the types of measures being detected, so highly intractable to search this large space X - haustively. • A Learning Classifier Scheme generates initial set of measures. • Refined using “Rule Evaluation” techniques like mutation and crossover. • Highly predictable sets of intrusions are combined and then searched, while lower predictable ones are weeded out. … Continued • Combined Belief in I is P ( I | A1, A2... An) = P (A1, A2... An | I ) x P ( I ) P(A1, A2... An ) • Each Ai depends only on I and is independent of the other measures of Ai , j = i, then n P (A1, A2... An | I ) = i=1P(Ai | I ) P (A1, A2... An | ~I ) = n P(Ai | ~I ) , and then i=1 P ( I | A1, A2... An) P( I ) x P ( ~I | A1, A2... An) P(~ I ) x n P(Ai | I ) i=1 n P(A |~ I ) i i=1 Inference • Determine odds of an intrusion given, – Values of various anomaly measure from the prior odds of intrusion, likelihood of each measure being anomalous is presence of INTRUSION !! A Trivial Bayesian Belief Network modeling Network Intrusion activity INTRUSION Too many users Too many CPU Intensive jobs Too many Disk Intensive jobs Trashing DISK I/O CPU Fragmentation Newly available program on the net Net I/O Conditional Probability • If we wish to detect P( Intrusion | Event Pattern) By Baye’s Law, P( Intrusion | Event Pattern) = P(Event Pattern | Intrusion ) x P(Intrusion) P(Event Pattern) P(Event Sequence) = (P( ES | I ) where, ES - Event Sequence, I - Intrusion. P( ES | ~I)) x P( I ) + P( ES | ~I) Web Security... • IN A NUTSHELL – A set of procedures, practices, and technology to protect web severs,web users and the surrounding Organizations. – Internet is a two way network. – Used by everybody. – Potential security flaws exist. – If subverted can be used by bad guys for malicious operations. – Unsophisticated users are the major population. – More expensive and time-consuming. Why worry about web security ? • Publicity - successful attack is a public event. • Commerce - Money is involved as transactions. • Proprietary Info - used to distributed information to internal members and external partners. • Network Access - used by people both inside and outside an organization. • Others - like server X-tensibility, Browser X-tensibility, disruption of service, complicated support, etc., The Web Security Problem... • Securing information in transit. – Digital Identification Techniques – Cryptography and Web security • Securing the Web Server. – Host and site security. (Firewalls) Digital Identification Techniques • Digital Signatures, – a protocol that produces the same effect as real signature. – Only the sender can mark it. – Easily identifiable by others as one from the sender. – Used to confirm agreement to a message. – Similar to biometrics such as image of a person’s face , fingerprints, Hand shape and size, DNA patterns, Voice prints, pattern of blood vessels in Retina, etc. Contd.. Digital Signatures must be – Authentic : Person B receives the pair [M, S(A,M)] from A, B can check if it came from A, and also this is firmly attached to M. – Unforgeable: Person A signs message M with sign S(A, M) , it is impossible for others to produce a pair S[M, S(A,M)]. Impostor A A Authentication Protects B Digital Signature C B Unforgeable Protects A Symmetric Key Digital Signatures with Arbiter – Private key encryption system guarantees authenticity of message as well as its secrecy. 1. S sends sealed M to Arbiter Arbiter Arbiter retrieves M Plain text M from S Ks M Sender Ks M S M Symmetric Key Digital Signature with Arbiter KR Recipient 3. Arbiter seals [S’s sealed M, identity of S, and copy of plain text M and sends all to R] Asymmetric Key Digital Signature E(D(M, KS), KR) D:KS M S Decrypts M For authenticity, D:KS unforgeability R Saves a copy to answer future disputes. M Certification Authorities(CA’s) – An organization that issues public key certificates(Digital Signature). – Certificates are synonymous to cryptographically signed index cards. – Signed by certification authority’s own private keys, contains name of the person, person’s public key, a serial number, and other info., – Example: verisign corp. Types of CA • Internal CA – can operate to certify its own employees – used to control access to internal resources • Outsourced Employee and Customer CA – leases trusted outsiders to certify employees and potential customers respectively. • Trusted Third Party CA – no prior relationship can establish and engage in legal transactions and business. Different kinds of certificates • Certification authorities Certificates – contain public key of CAs and name of service – this can in turn be signed by other certification authorities. • Server Certificates – contain public key of SSL server, – name of the organization running the server, Internet hostname, server’s public key. Contd... • Personal Certificates – contains individual’s name and public key. – other information is also allowed. • Software Publisher Certificates – certificates used to sign the distributed software. Cryptography and Web Security • Functions, – Confidentiality, • Encryption is used to scramble the message. – Authentication, • Digital Signatures are used for verification. – Integrity, • methods used to verify whether the message has been modified on transit. • Digital Signed message codes are used. – Nonrepudiation • author of a message can’t deny sending a message. What cryptography can’t do ? • • • • Protect unencrypted documents. Protect against stolen encryption keys. Against denial-of-service attacks. Against the record of a note that a message was sent. • Against a traitor or a mistake. Working Encryption Systems • Programs – PGP(Pretty Good Privacy). – S/MIME. • Protocols – – – – SSL(Secure Socket Layer). PCT(Private Communications Technology). S-HTTP(Secure HTTP). Cybercash. Contd… – SET(used in web shopping). • “Electronic Wallet” with User. • Server that runs on Merchant’s web site. • SET payment server runs in merchant’s bank. – DNSSEC(Domain Name System Security). – IPSec and IPv6. • IPsec works with IPv4 and standard version used today works for IPv6 and includes IPsec. – Kerberos. Network Layer Security Protocol (IPsec) • IP Security protocol - a suite of protocols that provides security at the network layer. • Network layer must provide – Secrecy - hide message from any third party that is "wire tapping" the network. – Source authentication -IP datagram with a particular IP source address, it might authenticate the source. • there are two principal protocols: – the Authentication Header (AH) protocol. provides source authentication and data integrity but not secrecy. – the Encapsulation Security Payload (ESP) protocol. provides data integrity and secrecy. Security Agreement (SA) - the source and network hosts handshake and create a network layer logical connection Authentication Header Protocol (AH) • SA is set up • source can send secured datagrams, that include the AH header. • Value of 51 in IP header means a AH Header has been included. • AH Headers have the following fields, – Next Header field, same as the protocol field for an ordinary datagram. – Security Parameter Index (SPI) field, an arbitrary 32-bit value that, in combination with the destination IP address and the security protocol, uniquely identifies the SA for the datagram. – Sequence Number field, a 32-bit field containing a sequence number – Authentication Data field, a variable-length field containing signed message digest (i.e., a digital signature) for this packet. Encapsulation Security Payload Protocol (ESP) • SA is set up. • source can send secured datagrams, that include the AH header. • a secured datagram is created by surrounding the original IP datagram data with header and trailer fields, • value 50 is used to indicate that the datagram includes an ESP header and trailer. • ESP header consists of a 32-bit field for the SPI and 32-bit field for the sequence number, which have exactly the same role as in the AH protocol. • trailer includes the Next Header field. What is SSL ? • Exists between raw TCP/IP and Application Layer. • Features added to streams by SSL – – – – Authentication and Nonrepudiation of Server, using Digital Signatures. Authentication and Nonrepudiation of Client, using Digital Signatures. Data confidentiality through Encryption. Data Integrity through the use of message authentication codes. • Functions – – – – Separation of duties. Efficiency. Certification - based authentication Protocol Agnostic. • Transport Layer Security is being tried out. Secure Web Server • Implements cryptographic protocols. • Safeguard any personal info received or collected. • Resistant to a determined attack over the I-net. SECURE WEB SERVER Bad Guys ATTACK ATTACK Bad Guys ATTACK Bad Guys SERVER ACTIVE AND PROVIDES SERVICES TO AUTHORIZED PERSONEL Web server Security • Host Security – Security of the computer on which web server is running. • Problems are due to – Security not viewed as a fundamental part of system setup and design. – Transmitting plain text and reusable password over networks. – Failure to obtain software to monitor bugs. – Failure to use security tools. – Lack of adequate Logging. – Lack of adequate backup procedures. – Lack of adequate System and network monitoring. – Failure to track security developments and take preventive action. Firewalls • Process that filters all traffic between a protected or “inside ” network and a less trustworthy or an “outside” network. • • • • Special form reference monitor. That which is not expressly forbidden is permitted. That which is expressly forbidden is not permitted. Challenge of protecting a network with a firewall is determining the security policy that meets the need of the installation. Types of Firewalls • Screening Router. Address 144.27.5.3 Allow in only A. Allow out only B , C. C A Address 100.24.4.0 Address 192.19.33.0 B Contd … • Route Screening outside Addresses 100.50.25.3 100.50.25.x 100.50.25.1 100.50.25.2 Subnet 100.50.25.x Screening Router Contd … • Proxy Gateway – Two headed piece of software. – runs pseudo applications. Logging WWW Access Address Local Area Network Remote Access Remote File Fetches Contd … • Guard – sophisticated proxy firewall. – Receives PDU’s interprets them passes the through same or different PDU’s. Proxy Firewall Address Screening Router Comparisons of Firewall types Screening Router Proxy Gateway Guard 1.Simplest. 2.Sees only address and service protocol type. 3.Auditing difficult. 4.Screens based on connection rules. 5.Complex addressing rules can make configuration tricky. 1.Somewhat complex. 2.Sees full text of communication. 3.Can audit activity. 4.Screens based on behavior of proxies. 5.Simple proxies can substitute for complex addressing rules 1.Most Complex. 2.Sees full text of communication. 3.Can audit activity. 4.Screens based on interpretation of message content. 5.Complex guard functionality can limit assurance. Encrypting Gateway (Virtual Private Network) • Internal encryption between offices for further protection. 144.27.5.3 Cryptographic Server 100.24.4.5 Address Proxy Gateway 100.24.4.0 192.19.33.0 Conclusions • Risks are involved in Computing. • Various techniques Encryption, Digital Signatures, Firewalls, etc can be used to provide security. • Web security is not a “Win” or “Loose” there is just a degree to which it can be realized. • No Solution is a complete solution !!! References • • • • Papers – “Model based Intrusion Detection” - Dorothy E Denning. – “Goal Oriented Auditing and Editing” - M.. Bishop. – “An Introduction to Intrusion Detection” - Aurobindo Sundaram. – “An application of Pattern Matching in Intrusion detection” - Sandeep Kumar and Eugene H Spafford. White Papers – “CERT Advisory CA-2000-02 Malicious Tags Embedded in Client Web Requests.” Books – “Security in Computing” - Charles P. Pfleeger. – “web Security and Commerce” - Simson Garfinkel and Gene Spafford. Websites – http://www.crypto.com/papers/ – http://www.awl.com/cseng/titles/0-201-63489-9/url_list.html – http://www.datafellows.com/products/white-papers/ – http://www.cerias.purdue.edu/coast/coast-library.html – http://www-net.cs.umass.edu/kurose/security/IPsec.htm – http://www.ietf.org/html.charters/tls-charter.html – http://www.counterpane.com Questions and Comments ???