Information Security Principles & Applications Topic 6: Security Policy Models 虞慧群 yhq@ecust.edu.cn Introduction Primary mission of information security is to ensure that systems and contents stay the same. If no threats, we could focus on improving systems, resulting in vast improvements in ease of use and usefulness. Attacks on information systems are a daily occurrence. Security Policy Defines what it means for a system to be secure Formally: Partitions a system into Set of secure (authorized) states Set of non-secure (unauthorized) states Secure system is one that Starts in authorized state Cannot enter unauthorized state Secure System - Example Unauthorized states A B C Authorized states Is this Finite State Machine Secure? A is start state ? B is start state ? C is start state ? How can this be made secure if not? Suppose A, B, and C are authorized states ? D Additional Definitions: Security breach: system enters an unauthorized state Let X be a set of entities, I be information. I has confidentiality with respect to X if no member of X can obtain information on I I has integrity with respect to X if all members of X trust I Trust I, its conveyance and protection (data integrity) I maybe origin information or an identity (authentication) I is a resource – its integrity implies it functions as it should (assurance) I has availability with respect to X if all members of X can access I Time limits (quality of service) Confidentiality Policy Also known as information flow policy Transfer of rights Transfer of information without transfer of rights Temporal context Highly developed in Military/Government Integrity Policy Defines how information can be altered Examples: Entities allowed to alter data Conditions under which data can be altered Limits to change of data Purchase over $1000 requires signature Check over $10,000 must be approved by one person and cashed by another Separation of duties : for preventing fraud Highly developed in commercial world Availability Policy An availability policy describes what services must be provided. It may present parameters within which the service will be accessible. It may require a level of service. Security Mechanism Policy describes what is allowed and/or what is not. Mechanism Example Policy: Students should not copy homework. An entity/procedure that enforces (part of) policy. Mechanism: Disallow access to files owned by other users. Does mechanism enforce policy? Security Model Security Policy: What is/isn’t authorized Problem: Policy specification often informal Implicit vs. Explicit Ambiguity Security Model: Model that represents a particular policy (policies) Model must be explicit, unambiguous Abstract details for analysis High-Level Policy Languages High-level: Independent of mechanisms Constraints expressed independent of enforcement mechanism Constraints restrict entities, actions Constraints expressed unambiguously Requires a precise language, usually a mathematical, logical, or programming-like language Example: Domain-Type Enforcement Language Subjects partitioned into domains Objects partitioned into types Each domain has a set of rights over each type Example: Web Browser Goal: restrict actions of Java programs that are downloaded and executed under control of web browser Language specific to Java programs Expresses constraints as conditions restricting invocation of entities Expressing Constraints Entities are classes, methods Operations Class: set of objects that an access constraint constrains Method: set of ways an operation can be invoked Instantiation: s creates instance of class c: s ├ c Invocation: s1 executes object s2: s1 |→s2 Access constraints deny(s op x) when b when b is true, subject s cannot perform op on (subject or class) x; empty s means all subjects Sample Constraints Downloaded program cannot access password database file on UNIX system Program’s class and methods for files: class File { public file(String name); public String getfilename(); public char read(); …. Constraint: deny(|→ file.read) when (file.getfilename() == “/etc/passwd”) Low-Level Policy Languages Low-level: close to mechanisms A set of inputs or arguments to commands that set, or check, constraints on a system Example: Tripwire: Flags what has changed Configuration file specifies settings to be checked History file keeps old (good) example Confidentiality Policy Also known as information flow policy Integrity is secondary objective Eg. Military mission date Bell-LaPadula Model Formally models military requirements Information has sensitivity levels or classification Subjects have clearance Subjects with clearance are allowed access Multi-level access control or mandatory access control Bell-LaPadula: Basics Mandatory access control Entities are assigned security levels Subject has security clearance L(s) = ls Object has security classification L(o) = lo Simplest case: Security levels are arranged in a linear order li < li+1 Example Top secret > Secret > Confidential >Unclassified “No Read Up” Information is allowed to flow up, not down Simple security property: - - s can read o if and only if lo ≤ ls and s has read access to o Combines mandatory (security levels) and discretionary (permission required) Prevents subjects from reading objects at higher levels (No Read Up rule) “No Write Down” Information is allowed to flow up, not down *property - - s can write o if and only if ls ≤ lo and s has write access to o Combines mandatory (security levels) and discretionary (permission required) Prevents subjects from writing to objects at lower levels (No Write Down rule) Example security level subject object Top Secret Tamara Personnel Files Secret Samuel E-Mail Files Confidential Claire Activity Logs Unclassified Ulaley Telephone Lists • Tamara can read which objects? And write? • Claire cannot read which objects? And write? • Ulaley can read which objects? And write? Access Rules Secure system: One in which both the properties hold Theorem: Let Σ be a system with secure initial state σ0, T be a set of state transformations If every element of T follows rules, every state σi secure Proof - induction Categories Total order of classifications not flexible enough Alice cleared for missiles; Bob cleared for warheads; Both cleared for targets Solution: Categories Use set of compartments (from power set of compartments) Enforce “need to know” principle Security levels (security level, category set) (Top Secret, {Nuc, Eur, Asi}) (Top Secret, {Nuc, Asi}) Combining with clearance: (L,C) dominates (L’,C’) L’ ≤ L and C’ C Induces lattice of security levels Lattice of categories {Nuc, Eur, Us} Examples of levels (Top Secret, {Nuc,Asi}) dom (Secret, {Nuc}) (Secret, {Nuc, Eur}) dom (Confidential, {Nuc,Eur}) (Top Secret, {Nuc}) dom (Confidential, {Eur}) {Nuc, Eur} {Nuc} {Nuc, Us} {Eur} Bounds Greatest lower, Lowest upper glb of {X, Nuc, Us} & {X, Eur, Us}? lub of {X, Nuc, Us} & {X, Eur, Us}? {} {Eur, Us} {Us} Access Rules Simple Security Condition: S can read O if and only if *-Property: S can write O if and only if S dominate O and S has read access to O O dom S and S has write access to O Secure system: One with above properties Theorem: Let Σ be a system with secure initial state σ0, T be a set of state transformations If every element of T follows rules, every state σi secure Problem: No write-down Cleared subject can’t communicate to non-cleared subject Any write from li to lk, i > k, would violate *-property Subject at li can only write to li and above Any read from lk to li, i > k, would violate simple security property Subject at lk can only read from lk and below Subject at level i can’t write something readable by subject at k Not very practical A solution: each subject has a maximum security level and a current security level. A subject may decrease its security level from maximum in order to communicate with entities at lower security levels. Integrity Policy Requirements 1. 2. 3. 4. 5. Commercial requirements differ from military requirements in their emphasis on preserving data integrity. Users will not write their own programs, but will use existing production programs and databases. Programmers will develop and test programs on a nonproduction system; if they need access to actual data, they will be given production data via a special process, but will use it on their development system. A special process must be followed to install a program from the development system onto the production system. The special process in requirement 3 must be controlled and audited. The managers and auditors must have access to both the system state and the system logs that are generated. Integrity Policy: Principles of operation These requirements induce principles of operation: Separation of Duty: Single person should not be allowed to carry out all steps of a critical function Moving a program from Dev. to Prod. system Developer and Certifier (installer) of a program Authorizing checks and cashing it Separation of function Do not process production data on development system Auditing Emphasis on recovery and accountability Controlled/audited process for updating code on production system Biba’s Integrity Policy Model Based on Bell-LaPadula (a mathematical dual of BL) Subject, Objects Integrity Levels with dominance relation Higher levels more reliable/trustworthy More accurate Information transfer path: Sequence of subjects, objects where s i r oi si w oi+1 Policies Low-Water-Mark Policy Ring Policy sro s w o i(o) ≤ i(s) s1 x s2 i(s2) ≤ i(s1) allows any subject to read any object (same as above) s r o i(s) ≤ i(o) s w o i(o) ≤ i(s) s1 x s2 i(s2) ≤ i(s1) (no read-down) (no write-up) Biba’s Model: Strict Integrity Policy (dual of Bell-LaPadula) s w o i(o) ≤ i(s) prevents writing to higher level s r o i’(s) = min(i(s), i(o)) drops subject’s level s1 x s2 i(s2) ≤ i(s1) prevents executing higher level objects Theorem for each: If there is an information transfer path from object o1 to object on+1, then the enforcement of the policy requires that i(on+1) ≤ i(o1) for all n>1 Chinese Wall Model Supports confidentiality and integrity, i.e. a hybrid policy Information can’t flow between items in a Conflict of Interest set Applicable to environment of stock exchange or investment house Models conflict of interest Objects: items of information related to a company Company dataset (CD): contains objects related to a single company Written CD(O) Conflict of interest class (COI): contains datasets of companies in competition Written COI(O) Assume: each object belongs to exactly one COI class Example Bank COI Class Bank of America Gasoline Company COI Class Shell Oil Standard Oil Union ’76 ARCO a Citibank Bank of the West a CW-Simple Security Property (Read rule) CW-Simple Security Property s can read o one of the following holds o’ PR(s) such that CD(o’) = CD(o) o’, o’ PR(s) COI(o’) COI(o), or o has been “sanitized” (o’ PR(s) indicates o’ has been previously read by s) Public information may belong to a CD As is publicly available, no conflicts of interest arise So, should not affect ability of analysts to read Typically, all sensitive data removed from such information before it is released publicly (called sanitization) Writing Anthony, Susan work in the same trading house Anthony can read BankOfAmercia’s CD, Susan can read Bank CitiBanks’s CD, Both can read ARCO’s CD If Anthony could write to Gas’ CD, Susan can read it Hence, indirectly, she can read information from BankOfAmercia’s CD, a clear conflict of interest CW-*-Property (Write rule) CW-*- Property s can write o both of the following conditions hold. The CW-simple security condition permits S to read O. For all unsanitized objects o’, s can read o’ CD(o’) = CD(o) Says that s can write to an object if all the (unsanitized) objects it can read are in the same dataset Anthony can read both CDs hence condition 1 is met He can read unsanitized objects of BankOfAmercia, hence condition 2 is false Hence Anthony can’t write to objects in ARCO’s CD. Role Based Access Control http://csrc.nist.gov/groups/SNS/rbac/ Access control in organizations is based on “roles that individual users take on as part of the organization” A role is “is a collection of permissions” RBAC Access depends on function, not identity Example: Allison is bookkeeper for Math Dept. She has access to financial records. If she leaves and Betty is hired as the new bookkeeper, Betty now has access to those records. The role of “bookkeeper” dictates access, not the identity of the individual. Advantages of RBAC Allows Efficient Security Management Administrative roles, Role hierarchy Principle of least privilege allows minimizing damage Separation of Duties constraints to prevent fraud Allows grouping of objects Policy-neutral - Provides generality Encompasses DAC and MAC policies RBAC Users Permission Users Permissions u1 o1 u1 o1 o2 u2 o2 om un om Manager Senior Administrator Senior Engineer Administrator Engineer Employee u2 Role r un n+m assignments nm assignments (a) (b) RBAC (NIST Standard) PA UA Users Roles Operations Objects Permissions user_sessions (one-to-many) role_sessions (many-to-many) Sessions An important difference from classical models is that Subject in other models corresponds to a Session in RBAC Core RBAC (relations) Permissions = 2Operations x Objects UA ⊆ Users x Roles PA ⊆ Permissions x Roles assigned_users: Roles 2Users assigned_permissions: Roles 2Permissions Op(p): set of operations associated with permission p Ob(p): set of objects associated with permission p user_sessions: Users 2Sessions session_user: Sessions Users session_roles: Sessions 2Roles session_roles(s) = {r | (session_user(s), r) UA)} avail_session_perms: Sessions 2Permissions RBAC with General Role Hierarchy RH (role hierarchy) PA UA Users Roles Operations Objects Permissions user_sessions (one-to-many) role_sessions (many-to-many) Sessions RBAC with General Role Hierarchy authorized_users: Roles 2Users authorized_users(r) = {u | r’ ≥ r &(r’, u) UA) authorized_permissions: Roles 2Permissions authorized_permissions (r) = {p | r’ ≥ r &(p, r’) PA) RH ⊆ Roles x Roles is a partial order called the inheritance relation written as ≥. (r1 ≥ r2) authorized_users(r1) ⊆ authorized_users(r2) & authorized_permisssions(r2) ⊆ authorized_permisssions(r1) Example Manager pxe, 5py Senior Administrator pa, pb e p3x, e p4y Senior Engineer px,x py e p1x, e p2y Administrator px, py Engineer Employee p1, p2 authorized_users(Employee)? authorized_users(Administrator)? authorized_permissions(Employee)? authorized_permissions(Administrator)? Constrained RBAC RH (role hierarchy) Static Separation of Duty PA UA Users Roles Operations Objects Permissions user_sessions (one-to-many) Sessions Dynamic Separation of Duty Static Separation of Duty SSD ⊆2Roles x N In absence of hierarchy Collection of pairs (RS, n) where RS is a role set, n ≥ 2; for all (RS, n) SSD, for all t ⊆RS: |t| ≥ n ∩rt assigned_users(r)= In presence of hierarchy Collection of pairs (RS, n) where RS is a role set, n ≥ 2; for all (RS, n) SSD, for all t ⊆RS: |t| ≥ n ∩rt authorized_uers(r)= Dynamic Separation of Duty DSD ⊆2Roles x N Collection of pairs (RS, n) where RS is a role set, n ≥ 2; A user cannot activate n or more roles from RS Formally?? [HW3?] What if both SSD and DSD contains (RS, n)? Consider (RS, n) = ({r1, r2, r3}, 2)? If SSD – can r1, r2 and r3 be assigned to u? If DSD – can r1, r2 and r3 be assigned to u? MAC using RBAC M1 H HR LW BLP Read Roles M1R (same lattice) M2R Write Roles M1W (inverse lattice) M2W LR HW L M2 Transformation rules • R = {L1R, L2R,…, LnR, L1W, L2W,…, LnW} • Two separate hierarchies for {L1R, L2R,…, LnR} and { L1W, L2W,…, LnW} • Each user is assigned to exactly two roles: xR and LW • Each session has exactly two roles yR and yW • Permission (o, r) is assigned to xR iff (o, w) is assigned to xW) RBAC’s Benefits Summary Policy describes what is allowed in a system. Confidentiality policies Integrity policies Bell-LaPadula model Biba’s model Hybrid policies Chinese Wall model Role-Based Access Control (RBAC) Model