GEOPRIV H. Schulzrinne Internet-Draft Columbia U. Expires: January 17, 2005 J. Morris CDT H. Tschofenig J. Cuellar Siemens J. Polk Cisco J. Rosenberg DynamicSoft July 19, 2004 A Document Format for Expressing Privacy Preferences draft-ietf-geopriv-common-policy-01.txt Status of this Memo By submitting this Internet-Draft, I certify that any applicable patent or other IPR claims of which I am aware have been disclosed, and any of which I become aware will be disclosed, in accordance with RFC 3668. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet-Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on January 17, 2005. Copyright Notice Copyright (C) The Internet Society (2004). All Rights Reserved. Abstract This document defines a framework for authorization policies Schulzrinne, et al. Expires January 17, 2005 [Page 1] Internet-Draft Common Policy July 2004 controling access to application specific data. This framework combines common location- and SIP-presence-specific authorization aspects. An XML schema specifies the language in which common policy rules are represented. The common policy framework can be extended to other application domains. Table of Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . 5 3. Modes of Operation . . . . . . . . . . . . . . . . . . . . . 6 3.1 Passive Request-Response - PS as Server (Responder) . . . 6 3.2 Active Request-Response - PS as Client (Initiator) . . . . 6 3.3 Event Notification . . . . . . . . . . . . . . . . . . . . 6 4. Goals and Assumptions . . . . . . . . . . . . . . . . . . . 8 5. Non-Goals . . . . . . . . . . . . . . . . . . . . . . . . . 10 6. Basic Data Model and Processing . . . . . . . . . . . . . . 11 6.1 Identification of Rules . . . . . . . . . . . . . . . . . 12 6.2 Extensions . . . . . . . . . . . . . . . . . . . . . . . . 12 7. Conditions . . . . . . . . . . . . . . . . . . . . . . . . . 13 7.1 Identity . . . . . . . . . . . . . . . . . . . . . . . . . 13 7.2 Sphere . . . . . . . . . . . . . . . . . . . . . . . . . . 14 7.3 Validity . . . . . . . . . . . . . . . . . . . . . . . . . 15 8. Actions . . . . . . . . . . . . . . . . . . . . . . . . . . 16 9. Transformations . . . . . . . . . . . . . . . . . . . . . . 17 10. Procedure for Combining Permissions . . . . . . . . . . . . 18 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . 18 10.2 Algorithm . . . . . . . . . . . . . . . . . . . . . . . 18 10.3 Example . . . . . . . . . . . . . . . . . . . . . . . . 19 11. Meta Policies . . . . . . . . . . . . . . . . . . . . . . . 22 12. Example . . . . . . . . . . . . . . . . . . . . . . . . . . 23 13. XML Schema Definition . . . . . . . . . . . . . . . . . . . 24 14. Security Considerations . . . . . . . . . . . . . . . . . . 27 15. IANA Considerations . . . . . . . . . . . . . . . . . . . . 28 15.1 Common Policy Namespace Registration . . . . . . . . . . 28 15.2 Common Policy Schema Registration . . . . . . . . . . . 28 16. References . . . . . . . . . . . . . . . . . . . . . . . . . 29 16.1 Normative References . . . . . . . . . . . . . . . . . . . 29 16.2 Informative References . . . . . . . . . . . . . . . . . . 29 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . 29 A. Contributors . . . . . . . . . . . . . . . . . . . . . . . . 31 B. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . 32 C. Enhancements to the Combining Permissions Algorithm . . . . 33 C.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 33 C.2 Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 36 C.3 Example . . . . . . . . . . . . . . . . . . . . . . . . . 37 Intellectual Property and Copyright Statements . . . . . . . 40 Schulzrinne, et al. Expires January 17, 2005 [Page 2] Internet-Draft Common Policy July 2004 1. Introduction This document defines a framework for creating authorization policies for access to application specific data. This framework is the result of combining the common aspects of single authorization systems that more specifically control access to presence and location information and that previously had been developed separately. The benefit of combining these two authorization systems is two-fold. First, it allows to build a system which enhances the value of presences with location information in a natural way and reuses the same underlying authorization mechanism. Second, it encourages a more generic authorization framework with mechanisms for extensibility. The applicability of the framework specified in this document is not limited to policies controling access to presence and location information data, but can be extended to other applications domains. The general framework defined in this document is intended to be accompanied and enhanced by application-specific policies specified elsewhere. Using the 'Location-specific Policy' and the 'Presence-specific Policy' documents [both are currently under development - references to be included here], figureFigure 1 illustrates the relationship between the 'Common Policy' framework defined in this document and application-specific enhancements of this framework. +-----------------+ | | | Common | | Policy | | | +---+---------+---+ /|\ /|\ | | +-------------------+ | | +-------------------+ | | | enhance | | | | Location-specific | | | | Presence-specific | | Policy |----+ +----| Policy | | | | | +-------------------+ +-------------------+ Figure 1: Common Policy Enhancements This document starts with an introduction to the terminology inSection 2, an illustration of basic modes of operation inSection 3, a description of goals (see Section 4) and non-goals (see Section 5) of the authorization policy framework, followed by the data model in Schulzrinne, et al. Expires January 17, 2005 [Page 3] Internet-Draft Common Policy July 2004 Section 6. The structure of a rule, namely conditions, actions and transformations, are described in Section 7, in Section 8 and in Section 9. The procedure for combining permissions is explained in Section 10 and used when more than one rule fires. An example is provided in Section 12. The XML schema will be discussed in Section 13. IANA considerations in Section 15 follow security considerations Section 14. Schulzrinne, et al. Expires January 17, 2005 [Page 4] Internet-Draft Common Policy July 2004 2. Terminology The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT","RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. This document introduces the following terms: PT - Presentity / Target: The PT is the entity about whom information has been requested. RM - Rule Maker: RM is an entity which creates the authorization rules which restrict access to data items. PS - (Authorization) Policy Server: This entity has access to both the authorization policies and to the data items. In location-specific applications, the entity PS is labeled as location server (LS). WR - Watcher / Recipient: This entity requests access to data items of the PT. An access operation might be either be a read, write or any other operation. In case of access to location information it might be a read operation. An 'authorization policy' is given by a 'rule set'. A 'rule set' contains an unordered list of 'rules'. A 'rule' has a 'conditions', an 'actions' and a 'transformations' part. The term 'permission' indicates the action and transformation components of a 'rule'. The terms 'authorization policy', 'policy' and 'rule set' are used interchangeable. The terms 'authorization policy rule', 'policy rule' and 'rule' are used interchangeable. The term 'using protocol' is defined in[RFC3693]. It refers to the protocol which is used to request access to and to return privacy sensitive data items. Schulzrinne, et al. Expires January 17, 2005 [Page 5] Internet-Draft Common Policy July 2004 3. Modes of Operation The abstract sequence of operations can roughly be described as follows. The PS receives a query for data items for a particular PT, via the using protocol. The using protocol provides the identity of the requestor (or more precisely the authentication protocol), either at the time of the query or at the subscription time. The authenticated identity of the WR, together with other information provided by the using protocol or generally available to the server, is then used for searching through the rule set. All matching rules are combined according to a permission combining algorithm described in Section 10. The result is returned to the WR, possibly modified by transformation policies. A single PS may authorize access to data items in more than one mode. Rather than having different rule sets for different modes all three modes are supported with a one rule set schema. Specific instances of the rule set can omit elements that are only applicable to the subscription model. The three different modes are explained below. 3.1 Passive Request-Response - PS as Server (Responder) In a passive request-response scenario, the WR queries the PS for data items about the PT. Examples of protocols following this mode of operation include HTTP, FTP, LDAP, finger or various RPC protocols, including Sun RPC, DCE, DCOM, Corba and SOAP. The PS uses the ruleset to determine whether the WR is authorized to access the PTs information, refusing the request if necessary. Furthermore, the PS might filter information by removing elements or by reducing the resolution of elements. 3.2 Active Request-Response - PS as Client (Initiator) Alternatively, the PS may contact the WR and convey data items. Examples include HTTP, SIP session setup (INVITE request), H.323 session setup or SMTP. 3.3 Event Notification Event notification adds a subscription phase to the "PS as client" mode of operation. A watcher or subscriber asks to be added to the notification list for a particular presentity or event. When the presentity changes state or the event occurs, the PS sends a message to the WR containing the updated state. (Presence is a special case of event notification; thus, we often use the term interchangeably.) In addition, the subscriber may itself add a filter to the subscription, limiting the rate or content of the notifications. If Schulzrinne, et al. Expires January 17, 2005 [Page 6] Internet-Draft Common Policy July 2004 an event, after filtering by the rulemaker-provided rules and by the subscriber-provided rules, only produces the same notification content that was sent previously, no event notification is sent. Schulzrinne, et al. Expires January 17, 2005 [Page 7] Internet-Draft Common Policy July 2004 4. Goals and Assumptions Below, we summarize our design goals and constraints. Table representation: Each rule must be representable as a row in a relational database. This design goal should allow efficient policy rule implementation by utilizing standard database optimization techniques. Permit only: Rules only provide permissions rather than denying them. Allowing both 'permit' and 'deny' actions would require some rule ordering which had implications on the update operations executed on these rules. Additionally it would make distributed rule sets more complicated. Hence, only 'permit' actions are allowed which result in more efficient rule processing. This also implies that rule ordering is not important. Consequently, to make a policy decision requires processing all policy rules. Additive permissions: A query for access to data items is matched against the rules in the rule database. If several rules match, then the overall permissions granted to the WR are the union of those permissions. A more detailed discussion is provided inSection 10. Upgradeable: It should be possible to add additional rules later, without breaking PSs that have not been upgraded. Any such upgrades must not degrade privacy constraints, but PSs not yet upgraded may reveal less information than the rulemaker would have chosen. Versioning support: In addition to the previous goal, a RM should be able to determine which types of rules are supported by the PS. The mechanism used to determine the capability of a PS will be covered in future versions of the document. Protocol-independent: The rule set supports constraints on both notifications or queries as well as subscriptions for event-based systems such as presence systems. No false assurance: It appears more dangerous to give the user the impression that the system will prevent disclosure automatically, but fail to do so with a significant probability of operator error or misunderstanding, than to force the user to explicitly invoke simpler rules. For example, rules based on weekday and time-of-day ranges seem particularly subject to misinterpretation and false assumptions on part of the RM. (For example, a non-technical RM would probably assume that the rules are based on the timezone of his current location, which may not be known to Schulzrinne, et al. Expires January 17, 2005 [Page 8] Internet-Draft Common Policy July 2004 other components of the system.) Schulzrinne, et al. Expires January 17, 2005 [Page 9] Internet-Draft Common Policy July 2004 5. Non-Goals We explicitly decided that a number of possibly worthwhile capabilities are beyond the scope of this first version. Future versions may include these capabilities, using the extension mechanism described in this document. Non-goals include: No external references: Attributes within specific rules cannot refer to external rule sets, databases, directories or other network elements. Any such external reference would make simple database implementation difficult and hence they are not supported in this version. No regular expression or wildcard matching: Conditions are matched on equality or 'greater-than'-style comparisons, not regular expressions, partial matches such as the SQL LIKE operator (e.g., LIKE "%foo%") or glob-style matches ("*@example.com"). Most of these are better expressed as explicit elements. No all-except conditions: It is not possible to express exclusion conditions based on identities such as "everybody except Alice". However, this restriction does not prevent all forms of blacklisting. It is still possible to express an authorization rule like 'I allow access to my location information for everyone of domain example.com except for John'. See the example in Section 7.1 describing how exceptions can be made work. No repeat times: Repeat times are difficult to make work correctly, due to the different time zones that PT, WR, PS and RM may occupy. It appears that suggestions for including time intervals are often based on supporting work/non-work distinctions, which unfortunately are difficult to capture by time alone. Schulzrinne, et al. Expires January 17, 2005 [Page 10] Internet-Draft Common Policy July 2004 6. Basic Data Model and Processing A rule set (or synonymously, a policy) consists of zero or more rules. The ordering of these rules is irrelevant. The rule set can be stored at the PS and conveyed from RM to PS as a single document, in subsets or as individual rules. A rule consists of three parts - conditions (seeSection 7), actions (see Section 8), and transformations (see Section 9). The conditions part is a set of expressions, each of which evaluates to either TRUE or FALSE, i.e. each of which is equipped with a value of either TRUE or FALSE by the PS. When a WR asks for information about a PT, the PS goes through each rule in the rule set. For each rule, it evaluates the expressions in the conditions part. If all of the expressions evaluate to TRUE, then the rule is applicable to this request. Generally, each expression specifies a condition based on some variable that is associated with the context of the request. These variables can include the identity of the WR, the domain of the WR, the time of day, or even external variables, such as the temperature or the mood of the PT. Assuming that the rule is applicable to the request, the actions and transformations (commonly referred to as permissions) in the rule specify how the PS is supposed to handle this request. If the request is to view the location of the PT, or to view its presence, the typical action is "permit", which allows the request to proceed. Assuming the action allows the request to proceed, the transformations part of the rule specifies how the information about the PT - their location information, their presence, etc. - is modified before being presented to the WR. These transformations are in the form of positive permissions. That is, they always specify a piece of information which is allowed to be seen by the WR. When a PS processes a request, it takes the transformations specified across all rules that match, and creates the union of them. The means for computing this union depend on the data type - Integer, Boolean, Set, or the Undef data type - and are described in more detail in Section 10. The resulting union effectively represents a "mask" - it defines what information is exposed to the WR. This mask is applied to the actual location or presence data for the PT, and the data which is permitted by the mask is shown to the WR. If the WR request a subset of information only (such as city-level civil location data only, instead of the full civil location information), the information delivered to the WR SHOULD be the intersection of the permissions granted to the WR and the data requested by the WR. In accordance to this document, rules are encoded in XML. To this end, Section 13 contains an XML schema defining the Common Policy Schulzrinne, et al. Expires January 17, 2005 [Page 11] Internet-Draft Common Policy July 2004 Markup Language. This, however, is purely an exchange format between RM and PS. The format does not imply that the RM or the PS use this format internally, e.g., in matching a query with the policy rules. The rules are designed so that a PS may translate the rules into a relational database table, with each rule represented by one row in the database. The database representation is by no means mandatory; we will use it as a convenient and widely-understood example of an internal representation. The database model has the advantage that operations on rows have tightly defined meanings. In addition, it appears plausible that larger-scale implementations will employ a backend database to store and query rules, as they can then benefit from existing optimized indexing, access control, scaling and integrity constraint mechanisms. Smaller-scale implementations may well choose different implementations, e.g., a simple traversal of the set of rules. 6.1 Identification of Rules Each rule is equipped with a parameter that identifies the rule. This rule identifier is an opaque token chosen by the RM. A RM MUST NOT use the same identifier for two rules that are available to the PS at the same time for a given PT. The combination uniquely identifies a rule. 6.2 Extensions The authorization policy framework defined in this document is meant to be extensible towards specific application domains. Such an extension is accomplished by defining conditions, actions and transformations that are specific to the desired application domain. Each extension MUST define its own namespace and indicate its version number. Schulzrinne, et al. Expires January 17, 2005 [Page 12] Internet-Draft Common Policy July 2004 7. Conditions The access to data items needs to be matched with the rule set stored at the PS. Each instance of a request has different attributes (e.g., the identity of the requestor) which are used for authorization. A rule in a rule set might have a number of conditions which need to be verified before executing the remaining parts of a rule (i.e., actions and transformations). Details about rule matching are described inSection 10. This document specifies only a few conditions (namely identity, sphere, and validity). Other conditions are left for extensions of this document. 7.1 Identity The policy framework specified in this document supports the usage of authenticated identities as input to access authorization decision processes. This framework, however, abstracts from the particularities of concrete authentication mechanisms employed by different using protocols and is therefore unable to specify explicitly the details of identity relevant information. Documents that enhance this framework should describe how a particular using protocol is able to provide identity information in a meaningful way. Such an enhancement needs to map the identity used by the authentication protocol employed in the using protocol to an identity used in the authorization policy. It is necessary to clearly define a mapping between the authenticated identity of the user (and the domain of the user) and the identities used in the authorization policies. This mapping needs to consider the large number of possible identities used in various authentication protocols and also to consider identities in using protocols. Furthermore, it is important to designate an identifier that denotes an 'anonymous user', i.e., a user that has not authenticated itself to the PS. The authors suggest to treat anonymous users by omitting this attribute in the rule which causes a 'NULL' value to be created in the ruleset table of a relational database. Any request for a data item (for a given PT) would match with respect to this attribute in a rule. Furthermore, pseudonyms need to be addressed as part of this mapping process. This specification provides an element which belongs to the group of condition elements. It can have either the or the element as child elements. The element contains a list of elements and allows to implement a simple blacklist mechanism. The element contains the identity without the domain part since it equals the domain of the element. The following example illustrates conditions based on an identity. Schulzrinne, et al. Expires January 17, 2005 [Page 13] Internet-Draft Common Policy July 2004 jack@example.com It is allowed to list more than one identity within a single rule as described in the following example. If multiple identities are provided in a single rule than the rule matches if one of the listed identities in a rule matches the authenticated identity of the entity requesting access to a resource. For the given example the rule matches if the entity requesting access to a resource is either alice@example.com or bob@example.com. alice@example.com bob@example.com The next example shows how exceptions are implemented. A request MUST match the domain part and all three exceptions parts in an atomic fashion to be a successful match. example.com joe tony mike 7.2 Sphere The element belongs to the group of condition elements. It can be used to indicate a state (e.g., 'work', 'home', 'meeting', 'travel') the PT is currently in. A sphere condition matches only if the PT is currently in the state indicated. The state may be conveyed by manual configuration or by some protocol. For example, RPID [I-D.ietf-impp-cpim-pidf] provides the ability to inform the PS of its current sphere. Switching from one sphere to another causes to switch between different modes of visibility. As a result different subsets of rules might be applicable. An example of a rule fragment is shown below: Schulzrinne, et al. Expires January 17, 2005 [Page 14] Internet-Draft Common Policy July 2004 work andrew@example.com home allison@example.com The code snippet above illustrates that the rule with the entity andrew@example.com matches if the sphere is been set to 'work'. In the second rule with the entity allison@example.com matches if the sphere is set to 'home'. 7.3 Validity The element is the third condition element specified in this document. It expresses the rule validity period by two attributes, a starting and a ending time. Times are expressed in XML dateTime format. Expressing the lifetime of a rule implements a garbage collection mechanism. A rule maker might not have always access to the PS to remove some rules which grant permissions. Hence this mechanisms allows to remove or invalidate granted permissions automatically without further interaction between the rule maker and the PS. An example of a rule fragment is shown below: 2003-08-15T10:20:00.000-05:00 2003-09-15T10:20:00.000-05:00 The , the and the element MUST NOT appear more than one in the conditions part of a single rule. The attribute discussed in Section 7 is 'sphere', its data type is 'string', and its value may be set to 'home'. The values of attributes of the same name MUST all be of the same data type. To evaluate a condition means to associate either TRUE or FALSE to the condition. A rule matches if all conditions contained in the conditions part of a rule evaluate to TRUE. When the PS receives a request for access to privacy-sensitive data then it needs to be matched against a rule set. The conditions part of each individual rule is evaluated and as a result one or more rules might match. If only a single rule matches then the result is determined by executing the actions and the transformations part following the conditions part of a rule. However, it can also be the case that two or more matching rules contain a permission of the same name (e.g., two rules contain a permission named 'precision of geospatial location information'), but do not specify the same value for that permission (e.g., the two rule might specify values of '10 km' and '200 km', respectively, for the permission named 'precision of geospatial location information'). This section describes the procedure for combining permissions in such cases. The values of attributes MUST be of either Boolean, Integer, Set or undefined. The value is undefined if no value is given for a particular attribute. Attributes with values of data type Integer can also be used for enumerations. For example, you can enumerate different levels of civil location information precision (e.g., level 0 "country, city, street" vs. level 1 "country, city") by associating integers to these levels. 10.2 Algorithm This section describes the algorithm in a more formal fashion. Schulzrinne, et al. Expires January 17, 2005 [Page 18] Internet-Draft Common Policy July 2004 The combining rules are simple and depend on the data types of the values of permissions: Let P be a policy. Let M be the subset of P consisting of rules r in P that match with respect to a given request. Let n be a name of a permission contained in a rule r in M, and let M(n) be the subset of M consisting of rules r in M that have a permission of name n. For each rule r in M(n), let v(r,n) and d(r,n) be the value and the data type, respectively, of the attribute of r with name n. Finally, let V(n) be the combined value of all the permissions values v(r,n), r in M(n). The combining rules that lead to the resulting value V(n) are the following: CR 1: If d(r,n)=Boolean or d(r,n)=Undefined for all r in M(n), then V(n) is given as follows: If there is a r in M(n) with v(r,n)=TRUE, then V(n)=TRUE. Otherwise, V(n)=FALSE. CR 2: If d(r,n)=Integer or d(r,n)=Undefined for all r in M(n), then V(n) is given as follows: If v(r,n)=undefined for all r in M(n), then V(n) is not specified by this specification. Otherwise, V(n)=max{v(r,n) | r in M(n)}. CR 3: If d(r,n)=Set or d(r,n)=Undefined for all r in M(n), then V(n) is given as follows: V(n)=union of all v(r,n), the union to be computed over all r in M(n) with v(r,n)!=undefined. 10.3 Example In the following example we illustrate the process of combining permissions. We will consider three conditions for our purpose, namely those of name identity, sphere, and validity. For editorial reasons the rule set in this example is represented in a table. Furthermore, the domain part of the identity of the WR is omitted. For actions we use two permissions with names X and Y. The values of X and Y are of data types Boolean and Integer, respectively. Permission X might, for example, represent the action. For transformations we use the attribute with the name Z whose value can be set either to '+'(or 1), 'o' (or 2) or '-' (or 3). Permission Z allows us to show the granularity reduction whereby a value of '+' shows the corresponding information unrestricted and '-' shows nothing. This permission might be related to location information or other presence attributes like mood. Internally we use the data type Integer for computing the permission of this attribute. Schulzrinne, et al. Expires January 17, 2005 [Page 19] Internet-Draft Common Policy July 2004 Conditions Actions/Transformations +--------------------------------+---------------------+ | Id WR-ID sphere from to | X Y Z | +--------------------------------+---------------------+ | 1 bob home A1 A2 | TRUE 10 o | | 2 alice work A1 A2 | FALSE 5 + | | 3 bob work A1 A2 | TRUE 3 - | | 4 tom work A1 A2 | TRUE 5 + | | 5 bob work A1 A3 | undef 12 o | | 6 bob work B1 B2 | FALSE 10 - | +--------------------------------+---------------------+ Again for editorial reasons, we use the following abbreviations for the two attributes 'from' and 'to': A1=2003-12-24T17:00:00+01:00 A2=2003-12-24T21:00:00+01:00 A3=2003-12-24T23:30:00+01:00 B1=2003-12-22T17:00:00+01:00 B2=2003-12-23T17:00:00+01:00 The entity 'bob' acts as a WR and requests data items. The policy P consists of the six rules shown in the table and identified by the values 1 to 6 in the 'Id' column. The PS receives the query at 2003-12-24T17:15:00+01:00. The value of the attribute with name 'sphere' indicating the state the PT is currently in is set to 'work'. Rule 1 does not match since the sphere condition does not match. Rule 2 does not match as the identity of the WR (here 'alice') does not equal 'bob'. Rule 3 matches since all conditions evaluate to TRUE. Rule 4 does not match as the identity of the WR (here 'tom') does not equal 'bob'. Rule 5 matches. Rule 6 does not match since the rule is not valid anymore. Therefore, the set M of matching rules consists of the rules 3 and 5. These two rules are used to compute the combined permission V(X), V(Y), and V(Z) for each of the permissions X, Y, and Z: Actions/Transformations +-----------------------+ | X Y Z | +-----------------------+ | TRUE 3 - | | undef 12 o | +-----------------------+ Schulzrinne, et al. Expires January 17, 2005 [Page 20] Internet-Draft Common Policy July 2004 The results of the permission combining algorithm is shown below. The combined value V(X) regarding the permission with name X equals TRUE according to the first combining rule listed above. The maximum of 3 and 12 is 12, so that V(Y)=12. For the attribute Z in this example the maximum between 'o' and '-' (i.e., between 2 and 3) is '-'. Actions/Transformations +-----------------------+ | X Y Z | +-----------------------+ | TRUE 12 - | +-----------------------+ Documents that extend the authorization policy framework defined here by introducing application specific actions and transformation MUST NOT define permissions whose values are of data type other than Boolean, Integer, Set, and Undef. Furthermore, permissions and the meaning of their values MUST be defined in such a way that the usage of the combining rules CR 1, CR2, and CR 3 always preserves or increases the level of privacy protection for the PT. In other words, the definition of new permissions MUST respect the way in which CR 1, CR 2, and CF 3 have been formulated in order to guarantee an appropriate level of privacy protection. Explicitly, it is not allowed to introduce a new permission whose value is of data type ... ... Boolean and the PS-side operation corresponding to the permission value TRUE has a lower privacy protection level than that operation that corresponds to the value FALSE. ... Integer and for any two permission values v1 and v2, v1 > v2, the PS-side operation corresponding to the value v1 has a lower privacy protection level than that operation that corresponds to the value v2. ... Set and for any two permission values s1 and s2, the PS-side operation corresponding to the union of s1 and s2 has a lower privacy protection level than those operations that correspond to s1 or s2. Schulzrinne, et al. Expires January 17, 2005 [Page 21] Internet-Draft Common Policy July 2004 11. Meta Policies Meta policies authorize a rulemaker to insert, update or delete a particular rule or an entire rule set. Some authorization policies are required to prevent unauthorized modification of rule sets. Meta policies are outside the scope of this document. A simple implementation could restrict access to the rule set only to the PT but more sophisticated mechanisms could be useful. As an example of such policies one could think of parents configuring the policies for their children. Schulzrinne, et al. Expires January 17, 2005 [Page 22] Internet-Draft Common Policy July 2004 12. Example This section gives a basic example of an XML document valid with respect to the XML schema defined in Section 13. Semantically richer examples can be found in documents which extend this schema with application domain specific data (e.g., location or presence information). bob@example.com 2003-12-24T17:00:00+01:00 2003-12-24T19:00:00+01:00 Schulzrinne, et al. Expires January 17, 2005 [Page 23] Internet-Draft Common Policy July 2004 13. XML Schema Definition This section provides the XML schema definition for the common policy markup language described in this document. Schulzrinne, et al. Expires January 17, 2005 [Page 24] Internet-Draft Common Policy July 2004 Although the XML schema does not require detailed explanations the following issues are worth mentioning: Each of the , , and (plural!) elements consists of zero or more child elements that belong to the substitution groups 'condition', 'action', and 'transformation', respectively. The respective heads of these substitution groups are the elements , , and (singular!). These elements cannot be used directly in an instance document since they are labeled as abstract. Schulzrinne, et al. Expires January 17, 2005 [Page 25] Internet-Draft Common Policy July 2004 XML schemas that extend this common policy schema by introducing new conditions, actions, and transformations MUST declare to which of these three substitution group the respective attribute belongs. These new attribute elements can then be used as immediate child elements of the , , and elements, depending on to which substitution group they belong. Schulzrinne, et al. Expires January 17, 2005 [Page 26] Internet-Draft Common Policy July 2004 14. Security Considerations This document describes a framework for authorization policy rules. This framework is intended to be enhanced elsewhere towards application domain specific data. Security considerations are to a great extent application data dependent, and therefore need to be covered by documents that extend the framework defined in this specification. However, new action and transformation permissions along with their allowed values must be defined in a way so that the usage of the permissions combining rules of Section 10 does not lower the level of privacy protection. See Section 10 for more details on this privacy issue. Schulzrinne, et al. Expires January 17, 2005 [Page 27] Internet-Draft Common Policy July 2004 15. IANA Considerations This section registers a new XML namespace and a new XML schema with IANA. 15.1 Common Policy Namespace Registration URI: urn:ietf:params:xml:ns:common-policy Registrant Contact: IETF Geopriv Working Group, Henning Schulzrinne (hgs+geopriv@cs.columbia.edu). XML: BEGIN Common Policy Namespace

Namespace for Common Authorization Policies

urn:ietf:params:xml:ns:common-policy

See RFCXXXX.

END 15.2 Common Policy Schema Registration URI: Please assign. Registrant Contact: IETF Geopriv Working Group, Henning Schulzrinne (hgs+geopriv@cs.columbia.edu). XML: The XML schema to be registered is contained in Section 13. Its first line is and its last line is Schulzrinne, et al. Expires January 17, 2005 [Page 28] Internet-Draft Common Policy July 2004 16. References 16.1 Normative References [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", March 1997. 16.2 Informative References [I-D.ietf-impp-cpim-pidf] Sugano, H., Fujimoto, S., Klyne, G., Bateman, A., Carr, W. and J. Peterson, "Presence Information Data Format (PIDF)", draft-ietf-impp-cpim-pidf-08 (work in progress), May 2003, . [RFC3693] Cuellar, J., Morris, J., Mulligan, D., Peterson, J. and J. Polk, "Geopriv Requirements", RFC 3693, February 2004, . Authors' Addresses Henning Schulzrinne Columbia University Department of Computer Science 450 Computer Science Building New York, NY 10027 USA Phone: +1 212 939 7042 EMail: schulzrinne@cs.columbia.edu URI: http://www.cs.columbia.edu/~hgs John B. Morris, Jr. Center for Democracy and Technology 1634 I Street NW, Suite 1100 Washington, DC 20006 USA EMail: jmorris@cdt.org URI: http://www.cdt.org Schulzrinne, et al. Expires January 17, 2005 [Page 29] Internet-Draft Common Policy July 2004 Hannes Tschofenig Siemens Otto-Hahn-Ring 6 Munich, Bayern 81739 Germany EMail: Hannes.Tschofenig@siemens.com URI: http://www.tschofenig.com Jorge R. Cuellar Siemens Otto-Hahn-Ring 6 Munich, Bayern 81739 Germany EMail: Jorge.Cuellar@siemens.com James Polk Cisco 2200 East President George Bush Turnpike Richardson, Texas 75082 USA EMail: jmpolk@cisco.com Jonathan Rosenberg DynamicSoft 600 Lanidex Plaza Parsippany, New York 07054 USA EMail: jdrosen@dynamicsoft.com URI: http://www.jdrosen.net Schulzrinne, et al. Expires January 17, 2005 [Page 30] Internet-Draft Common Policy July 2004 Appendix A. Contributors We would like to thank Christian Guenther for his help with this document. Christian Guenther Siemens AG Corporate Technology 81730 Munich Email: christian.guenther@siemens.com Germany Schulzrinne, et al. Expires January 17, 2005 [Page 31] Internet-Draft Common Policy July 2004 Appendix B. Acknowledgments This document is partially based on the discussions within the IETF GEOPRIV working group. Discussions at the Geopriv Interim Meeting 2003 in Washington, D.C., helped the working group to make progress on the authorization policies based on the discussions among the participants. We particularly want to thank Allison Mankin , Randall Gellens , Andrew Newton , Ted Hardie , Jon Peterson for discussing a number of details with us. They helped us to improve the quality of this document. Furthermore, we would like to thank the IETF SIMPLE working group for their discussions of J. Rosenberg's draft on XCAP authorization policies. We thank Stefan Berg, Christian Schmidt, Markus Isomaki and Eva Maria Leppanen for their comments. Schulzrinne, et al. Expires January 17, 2005 [Page 32] Internet-Draft Common Policy July 2004 Appendix C. Enhancements to the Combining Permissions Algorithm This section contains text, which should replace the text in Section 10, if approved by the Geopriv working group. It aims to enhance the combining permissions algorithm to offer a better privacy protection and to fix technical problems with the current algorithm. C.1 Introduction This section describes the mechanism that MUST be employed in order to determine the actions and transformations the PS has to employ when processing a request for privacy-sensitive data. When a PS receives such a request, the PS MUST evaluate the set of rules applicable to the request. According to this specification, a rule set consists of a set of rules each of which is composed of conditions, actions and transformations. First of all, the PS MUST determine the set of matching rules within the rule set. To this end, the PS MUST evaluate the conditions part of each rule contained in the rule set. To evaluate the conditions part of a rule means to associate either TRUE or FALSE to each condition contained in the conditions part of the rule. A rule matches if all conditions contained in the conditions part of a rule evaluate to TRUE. Secondly, the PS MUST determine the actions and transformations it has to perform. If the set of matching rules consists of a single rule only, then the PS MUST execute the actions and transformations as specified in that rule. However, it can also be the case that two or more matching rules contain a permission of the same name (e.g., two rules contain a permission named 'location-information-precision'), but do not specify the same value for that permission (e.g., the two rules might specify values of '10 km' and '200 km', respectively, for the permission named 'location-information-precision'). This section describes the procedure for combining actions and transformations in such cases. The notion 'permission' is used herein as a generic term for 'action' and 'transformation'. In order to come to an executable procedure for combining permissions, each permission definition MUST specify the name and the data type of it. Each permission MUST be either of data type 'Boolean', 'Integer', or 'Set'. In case of 'Boolean', the permission can have the values 'TRUE' or 'FALSE'. In case of 'Integer', the permitted values of a permission of this type are the integer number values. In case of 'Set', the definition of a permission of this type MUST also include the definition of the names and the elements of the sets that are permitted as values of such permissions. Schulzrinne, et al. Expires January 17, 2005 [Page 33] Internet-Draft Common Policy July 2004 For example, a permission named 'civil location information' that is to specify different precision levels of civil location information could be realized by specifying that it is of data type 'Set' and that the set named 'level 1' consists of the elements 'country', 'city', and 'street', while the set named 'level 2' has the elements 'country' and 'city' only. The data type 'Integer' also allows for enumerations when defining permissions: For example, instead of using the 'Set' data type, you could also define the permission indicating different precision levels of civil location information by enumerating these levels (e.g., level A, level B, ...), associating integer values to the single enumeration values (e.g., 1, 2, ...), and specifying the meaning of these values (e.g., level A stands for "country, city, street", level B stands for "country, city", and so on). This does not contradict the fact that permissions must be either of data type 'Boolean', 'Integer', or 'Set', as you MUST associate integer values to the single enumeration values when defining permission by enumeration. Now, it is necessary to have a means of preserving the level of privacy-protection when combining two permissions. The problem here is as follows: Assume, a first matching rule contains the permission named 'location-information-precision' of data type 'Integer' which has been equipped with the value 1 permitting the level of location information precision that consists of 'country', 'city' and 'street' data. A second matching rule (which might have a different set of conditions when compared to the first rule, but is also matching just like the first) also comprises the permission named 'location-information-precision', but this time specifying a value of 2 admitting 'country' and 'city' level precision only. In favor of privacy protection, it is necessary to combine these two permissions to a rule that permits 'country' and 'city' level location information only. Mathematically spoken and respecting the fact that the locatin-information-precision permission had been defined in this example in such a way that its values 1 and 2 represent lower and higher levels of pricacy protection, respectively, it is necessary to combine these two permissions by calculating the maximum of its single values: combined value = maximum of single values = max{1,2} = 2 = 'country' and 'city' level of precision. However, the 'location-information-precision' permission could have been specified in another way as well: the value of 1 could have been made representing 'country' and 'city' only, while the value of 2 could represent 'country', 'city' and 'street'. Now, it were necessary to combine these two rules by calculating the minimum and Schulzrinne, et al. Expires January 17, 2005 [Page 34] Internet-Draft Common Policy July 2004 not the maximum of the single value. Similar problems occur also for the permission data types 'Boolean' and 'Set'. Computer programs responsible for combining permissions must therefore get indicated which algorithm is to be employed when combining a permission of a given name. This is accomplished by the requirement that each permission definition has not only to specify its name and data type but also exactly one of the following six combining rules (CRs): CR-Boolean-Or CR-Boolean-And CR-Integer-Minimum CR-Integer-Maximum CR-Set-Intersection CR-Set-Union What these combining rules actually mean from the algorithmic perspective will be detailed in the next paragraph. From the XML point of view, authors of XML policy languages that are to be integrated under the roof of the common policy framework MUST use the XML built-in element when defining new permissions in XML schemas and equip this element with one of the six character strings above representing one of the six possible combining rules. This approach guarantees that each computer program that has access to the XML schema specifying a certain policy language within the common policy framework can apply the combining rule that is specified by the permission definition's child element . To give an example: A location information-specific XML policy language might define a permission named 'latitude-resolution' of data type 'Integer' which is to indicate the number of digits permitted to be sent to a certain set of receivers. In order to fulfill the requirement discussed above, the XML schema defining that policy language should specify the 'latitude-resolution' transformation as follows: Schulzrinne, et al. Expires January 17, 2005 [Page 35] Internet-Draft Common Policy July 2004 CR-Integer-Minimum C.2 Algorithms This section describes the algorithms for the six possible combining rules in a formal fashion. The combining rules are simple and depend on the data types of the values of permissions: Let P be a policy, i.e., a rule set. Let M be the subset of P consisting of rules r in P that match with respect to a given request. Let n be a name of a permission contained in a rule r in M, D(n) be the data type of the permissions of name n, and let M(n) be the subset of M consisting of rules r in M that have a permission of name n. For each rule r in M(n), let v(r,n) be the value of the permission of name n contained in the rule r. Finally, let CV(n) be the combined value of all the permissions values v(r,n), r in M(n). The combining rules CR-Boolean-Or, CR-Boolean-And, CR-Integer-Minimum, CR-Integer-Maximum, CR-Set-Intersection and CR-Set-Union MUST be implemented as follows: CR-Boolean-Or: This CR is applicable only if D(n)=Boolean: CV(n)=Or{v(r,n) | r in M(n)}. This means: CV(n)=TRUE if and only if there is a value v(r,n), r in M(n), with v(r,n)=TRUE. CR-Boolean-And: This CR is applicable only if D(n)=Boolean: CV(n)=And{v(r,n) | r in M(n)}. This means: CV(n)=TRUE if and only if for all r in M(n): v(r,n)=TRUE. CR-Integer-Minimum: This CR is applicable only if D(n)=Integer: CV(n)=Minimum{v(r,n) | r in M(n)}. CR-Integer-Maximum: This CR is applicable only if D(n)=Integer: CV(n)=Maximum{v(r,n) | r in M(n)}. CR-Set-Intersection: This CR is applicable only if D(n)=Set: CV(n)=Intersection{v(r,n) | r in M(n)}. Schulzrinne, et al. Expires January 17, 2005 [Page 36] Internet-Draft Common Policy July 2004 CR-Set-Union: This CR is applicable only if D(n)=Set: CV(n)=Union{v(r,n) | r in M(n)}. C.3 Example In the following example we illustrate the process of combining permissions. We will consider three conditions for our purpose, namely those of name identity, sphere, and validity. For editorial reasons the rule set in this example is represented in a table. Furthermore, the domain part of the identity of the WR is omitted. For actions we use two permissions with names X and Y. The values of X and Y are of data types Boolean and Integer, respectively. The combining rules that must be employed when combining values of X and Y are CR-Boolean-Or and CR-Integer-Maximum, respectively. For transformations we use the permission with the name Z whose value can be set either to '-'(or 1), 'o' (or 2) or '+' (or 3). Its combining rule is CR-Integer-Minimum. Permission Z allows us to show the granularity reduction whereby a value of '+' shows the corresponding information unrestricted and '-' shows nothing. This permission might be related to location information or other presence attributes like mood. Internally we use the data type Integer for computing the permission of this attribute. Conditions Actions/Transformations +--------------------------------+---------------------+ | Id WR-ID sphere from to | X Y Z | +--------------------------------+---------------------+ | 1 bob home A1 A2 | TRUE 10 o | | 2 alice work A1 A2 | FALSE 5 + | | 3 bob work A1 A2 | TRUE 3 - | | 4 tom work A1 A2 | TRUE 5 + | | 5 bob work A1 A3 | undef 12 o | | 6 bob work B1 B2 | FALSE 10 - | +--------------------------------+---------------------+ Again for editorial reasons, we use the following abbreviations for values of the two attributes 'from' and 'to': A1=2003-12-24T17:00:00+01:00 A2=2003-12-24T21:00:00+01:00 A3=2003-12-24T23:30:00+01:00 B1=2003-12-22T17:00:00+01:00 B2=2003-12-23T17:00:00+01:00 Schulzrinne, et al. Expires January 17, 2005 [Page 37] Internet-Draft Common Policy July 2004 The entity 'bob' acts as a WR and requests data items. The policy P consists of the six rules shown in the table and identified by the values 1 to 6 in the 'Id' column. The PS receives the query at 2003-12-24T17:15:00+01:00. The value of the attribute with name 'sphere' indicating the state the PT is currently in is set to 'work'. Rule 1 does not match since the sphere condition does not match. Rule 2 does not match as the identity of the WR (here 'alice') does not equal 'bob'. Rule 3 matches since all conditions evaluate to TRUE. Rule 4 does not match as the identity of the WR (here 'tom') does not equal 'bob'. Rule 5 matches. Rule 6 does not match since the rule is not valid anymore. Therefore, the set M of matching rules consists of the rules 3 and 5. These two rules are used to compute the combined permission values CV(X), CV(Y), and CV(Z) for each of the permissions X, Y, and Z: Actions/Transformations/Combining Rules +-----------------------------------------------------------+ | X Y Z | +-----------------------------------------------------------+ | TRUE 3 - (1) | | undef 12 o (2) | | CR-Boolean-Or CR-Integer-Maximum CR-Integer-Minimum | +-----------------------------------------------------------+ The results of the permission combining algorithms are shown below. The combined value CV(X) regarding the permission with name X equals TRUE according to its combining rule CR-Boolean-Or. The maximum of 3 and 12 is 12, so that CV(Y)=12. For the attribute Z in this example, the minimum between '-' and 'o' (i.e., between 1 and 2) is '-'. Actions/Transformations +-----------------------+ | X Y Z | +-----------------------+ | TRUE 12 - (1) | +-----------------------+ Documents that extend the authorization policy framework defined here by introducing application specific actions and transformations MUST NOT define permissions whose values are of data type other than Boolean, Integer, and Set. At least, the actual data type used in the permission definition MUST be representable by means of these three data types. In such cases, the mapping between the data type used and one of the three standard data types Boolean, Integer, and Schulzrinne, et al. Expires January 17, 2005 [Page 38] Internet-Draft Common Policy July 2004 Set MUST be given explicitly when using another permission data type (such as an enumeration data type or the data type that consists of the values '-', 'o' and '+' as illustrated above). Furthermore, permissions and the meaning of their values MUST be defined in such a way that the application of one of the six combining rules specified in this section actually preserves the level of privacy protection when determining combined values of single permission values contained in several matching rules. For example, this requirement implies that permission values of data type 'Integer' are ordered in such a way that either lower values correspond to lower privacy protection levels and higher values to higher levels, or vice versa. However, the value 1 MUST NOT correspond to a medium level of privacy protection, 3 to a lower and 2 to a higher, for instance, so that neither the application of CR-Integer-Minimum nor of CR-Integer-Maximum would result in reasonable, privacy-protecting combined value. Schulzrinne, et al. Expires January 17, 2005 [Page 39] Internet-Draft Common Policy July 2004 Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf-ipr@ietf.org. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Copyright Statement Copyright (C) The Internet Society (2004). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society. Schulzrinne, et al. Expires January 17, 2005 [Page 40]