(4) Identification

• Identity is required for any non-anonymous communications
  • groups can have an identity (facebook members see more than non-members)
  • pseudonyms are hidden identities (the real identity is not visible)
  • personal identity should be tied to a person itself
• Proof of Identity is important for any privileged operation
  • signatures and seals are traditional ways
  • traditional ways are mostly protected by law (but not really safe)
  • more modern ways often include technical methods for Authentication [Authentication (1)]
• Client identity on the Web can be bound in three ways:
  1. Computer (most of the time identified by an IP Address [Internet Architecture; IP Address (1)])
  2. Browser (in the form of a stored cookie [Web Storage; Cookies (1)])
  3. User (identified through some authentication method [Authentication (1)])

(5) Authentication

• Authentication is the process of verifying an identity
  • the weakest form of authentication is simply trust
  • legal consequences can make it more risky to falsify authentication
  • technical measures should make it hard to impossible to falsify authentication
• Authentication on the Web comes in many different flavors
  • implicitly by accessing a server from some IP Address [Internet Architecture; IP Address (1)] range
  • presenting a cookie [Web Storage; Cookies (1)] from a previous formal authentication
  • presenting a password as a proof of identity
  • proving that you are owning additional authentication hardware (often PIN [http://en.wikipedia.org/wiki/Personal_identification_number]-enabled)
• Risk and potential damage should justify authentication methods

(6) Authorization

• Authorization is the question of allowing operations
  • Identification [Identification (1)] is necessary to identify the initiator
  • Authentication [Authentication (1)] is necessary to verify the initiator's identity
  • if the initiator is authorized, the operation can be performed
• Web pages often are public or restricted access
  • public web pages do not require any identification (and thus authentication)
  • restricted access Web pages can be group pages (internal company pages)
  • personal access is another popular scenario (email, facebook, online banking)
• Web servers have well-defined ways of performing authentication [Hypertext Transfer Protocol (HTTP); HTTP Authentication (1)]

(8) Trust and Security on the Web

• Web-based applications introduce many risks
  • do you trust your browser? (it may not safeguard your information)
  • do you trust your computer? (it may have a virus)
  • do you trust your network? (it may be monitored on various levels)
  • do you trust the server? (it may be a fake phishing [http://en.wikipedia.org/wiki/Phishing] server)
• Most of these risks are amplified by the Web's scale
  • phishing and spamming only work because the Web makes fraud more effective
• Controlling Web access is important for safe browsing
  • trusting shared browsers is risky (they may store logins and cache pages)
  • trusting the network can be risky (more and more networks are wire-tapped)
  • trusting the server is risky (phishing and poor server security)

(9) Privacy Options

(10) Security Options

(11) Encryption Options

(13) Cryptography

• Cryptography is structured into different layers
  • layering is a well-established principle for separation of concerns
• Cryptographic primitives implement very basic functionality
  • changes and advancements in this area are limited to very specialized researchers
  • it is easy to make fatal mistakes which then challenge everything built on top if it
• Cryptographic protocols assemble primitives into application-level solutions
  • primitives solve very basic security problems (fingerprints, encryption, …)
  • protocols combine these into applications (digital signatures, secure communications, …)

(14) One-Way Function

• Hashes (or message digests) are well-known in computer science
• One-way functions are cryptographically safe hashes
  • very hard to find an input producing a given output
  • very hard to find two inputs producing the same output (collision)

(25) Secure Communications

• Public-Key cryptography [Public-Key Cryptography (1)] is computationally expensive
  • it is possible to encrypt all traffic using asymmetric key pairs
  • this generates considerably more load on the server side
• Combining public-key [Public-Key Cryptography (1)] and secret-key [Secret-Key Cryptography (1)] cryptography
  1. check the public key for authenticity (using a Certificate [Certificate (1)])
  2. generate a key for a secret-key encryption scheme
  3. use the public key to securely transmit the secret key
  4. use the secret key for securely transmitting the payload
• Combines the advantages of both methods
  • the lower complexity of secret-key algorithms
  • the ability of public-key algorithms to work without a secure channel

(26) HTTP and Security

• HTTP sends clear-text messages
• Making HTTP secure requires additional mechanisms
• Encryption is done by a layer on top of TCP
  • Secure Sockets Layer (SSL) is the protocol layer invented by Netscape
  • Transport Layer Security (TLS) is the standardized Internet version
  • TLS adds more encryption schemes and more flexibility
• Lower-level methods may also provide encryption
  • Virtual Private Networks (VPN) provide IP-based encryption
  • WEP and WPA provide network interface encryption

(27) HTTP and SSL

(29) Internet Security

• Security is hard
• Certificates are used to guarantee a party's authenticity
• Certificates are digital signatures issued by trusted parties
• One authenticated, public keys can be used to securely communicate
• Encryption on the Web is based on HTTPS