• Operationally Equivalent Translations

This is one of the possible Use Cases.

1. Abstract

If it is assumed that the same standard rule language and inference procedure is used everywhere then rule-sets could simply be exchanged without any need for a formal rule-translation process. This use case does not make that assumption. Therefore, as this use case exemplifies, useful “rule interchange” implies some notion of “equivalence” across different rule representation languages and various inference procedures. This use case is intended to explore the consequences of the idea that the RIF will be employed to facilitate translation across rule platforms.

2. Status

• Originally proposed by: Proposed to the First F2F WG meeting as Use Case MITRE-1 by AllenGinsberg.

• If based on existing document, project, system, etc., commercially or institutionally sanctioned or supported, then briefly describe that relationship

  • The notion of allowing a policy-driven approach to dynamic spectrum access is both a commercial and institutionally supported goal. See for example the decision by the European Union to allow "Dynamic Frequency Selection" (DFS) use of the 5 GHz frequency band intermittently used by military and weather radar) by wireless systems (http://europa.eu.int/eur-lex/lex/LexUriServ/site/en/oj/2005/l_187/l_18720050719en00220024.pdf).

• If based on perceived business or organizational need that is currently not addressed, then briefly describe potential commercial or institutional interest

  • Device manufacturers want to make devices that can take of advantage of dynamic spectrum access, but the real possibility of multiple policies written for different policy engines would be a problem. A means of providing operationally equivalent translations of policies for different devices would remove this potential bottleneck.

• Is any implementation effort underway?

  • Efforts to develop rule-based "policy engines" as well as policy languages and frameworks are underway. In particular, see the XG project at SRI International, http://www.ai.sri.com/project/XG-policy. A research project involving the use of semantic technologies, including OWL and rules, is underway at Mitre. See http://www.mitre.org/news/events/tech05/briefings/2153.pdf for more information.

3. Links to Related Use Cases

• Rule-based Service Level Agreements (SLA) and Web Services: Important similarity is the need or opportunity for third-party based service to mediate or enable rule interchange.

• Rule Based Service Level Management and SLAs for Service Oriented Computing: Same as above

• Message Transformation: Same as above.

4. Relationship to OWL/RDF Compatibility

Explain relationship to OWL/RDF Compatiblity issues

5. Examples of Rule Platforms Supporting this Use Case

• SWI Prolog: This is the language that the SRI effort is using to implement its XG policy language.

6. Benefits of Interchange

• Benefit 1: Enables “write once, run everywhere” paradigm.

• Benefit 2: Helps to determine equivalence or extent of differences among rule-sets written in different languages.

• Benefit 3: Enables “graceful degradation” by providing partial translations of a rule-set suitable for devices with restricted capabilities.

7. Requirements on the RIF

• Requirement 1: In addition to providing a rule interchange format, the RIF must also provide

  • A standard format for specifying the inference procedure (inference engine) that can be applied to a set of rules. For the sake of brevity we refer to this as the IPIF (Inference Procedure Interchange Format).
  • A standard format for specifying the intended interpretation (semantics) of a set of RIF rules with accompanying IPIF inference procedure. For the sake of brevity we refer to this as the SIF (Semantics Interchange Format). Note that the SIF is where a formal characterization of the possible inputs to a rule-set is provided.

• Requirement 2: Given a rule-set R written in the RIF together with an inference procedure P specified in the IPIF and an intended interpretation S specified in the SIM, it must be possible to construct a virtual machine that uses P to execute R for any input allowed by S.

8. Breakdown

8.1. Actors and their Goals

• Cognitive-Radio-User The user of the device wants it to work well and legally in as many locales as possible.

• Cognitive-Radio-Manufacturer The manufacturer of the device wants it to work well and legally in as many locales as possible and wants to support that goal without undue cost.

• Industry-Consortium-Members Want to make it easier and less costly for manufacturers to make devices that will work in different policy domains.

• Governments Want to write and enforce policies governing the use of spectrum in their jurisdictions in whatever way they see fit.

• International-Telecommunication-Union Wants to “ensure rational, equitable, efficient and economical use of the radio-frequency spectrum by all radio communication services.”

8.2. Main Sequence

• Step 0 The ITU-Radio (International Telecommunication Union – Radio Communication Sector) writes/updates a machine-readable version of the current internationally agreed upon regulations governing spectrum usage. (Step 0 might or might not precede Step 1.) We may assume that these regulations are written using the RIF and that IPIF and SIM specifications are also provided (see Requirement 1).

  Step 1 A national government writes/updates a machine-interpretable policy for the legal use of the electromagnetic spectrum resource in its jurisdiction.

  Step 2 The national policy is checked against the international regulations for any inconsistencies or other potential problems. This is done by translating the national policy into the RIF+IFIP+SIM in such a way that it can be merged with the ITU-Radio regulations. The resulting formal specification is then run on a virtual machine that tests for desired output using test cases drawn from the combined SIM (see Requirement 2). (The nation in question ultimately decides what to put in its policy.)

  Step 3 The national policy is published and deployed.

  Step 4 The industry consortium receives the national policy (in RIF+IFIP+SIM). For each device covered by the interests of the consortium, a machine-readable translation of the policy is created in a format useable by the device. That is, the RIF+IFIP+SIM version of the national policy is translated into multiple formats preserving operational equivalence, i.e., the observable behavior of devices or systems using any of these translations should be identical to that of the original for all possible inputs (see Requirement 2).

  Step 5 These translations are deployed.

  Step 6 A user of one of these devices crosses a national boundary. The device automatically detects the change in policy domain and contacts a consortium maintained server to download a version of the policy that it can use to operate legally in this country. The download takes places seamlessly.

8.3. Alternate Sequences

8.3.1. Policy not Available in necessary format; Use restricted alternate.

• Step 0 – Step 5 Same as main sequence.

  Step 6 A user of one of these devices crosses a national boundary. This time a translation of the necessary policy is not available in a format that device can use. However it is possible for the device to operate with restricted functionality by downloading a consortium-sanctioned policy that will run on the device.

  Step 7 The device informs its user about the situation. The user agrees to continue using the device with the alternate policy which is now downloaded.

9. Narratives

9.1. Supporting Global Policy Deployment

An international consortium of manufacturers wants to support the ability of intelligent wireless devices, “cognitive radios,” to operate across international boundaries. The goal is to develop a global policy deployment service that cognitive radios can access when attempting to operate in a non-native policy domain. For each type of cognitive radio device covered by the consortium there is a model of the cognitive and operational capabilities of the device. In conjunction with a policy and a “policy engine,” the device model essentially governs the behavior of the device. In principle, by “swapping in” a new policy the device can operate legally in any policy domain. But there are two bottlenecks: 1) the various national policies are written in different rule-platforms and 2) some device models are not capable of supporting all the language features required by the (union of) the policies.

10. Commentary

The MITRE-1 use case scenario attempts to explore the consequences of using the RIF for translation that attempts to preserve operational equivalence. This form of equivalence is weaker than logical equivalence. The latter puts constraints on everything that can be concluded with a rule-set, while the former is restricted to conclusions that involve operational or observable terms used in an application domain. This use case describes using a test-set approach for determining operational equivalence. Clearly a more efficient and theoretically satisfying approach is desirable. That should be possible, at least in some cases, by using information contained in 1) RIF+IFIP+SIM specifications, and 2) theoretical knowledge of the properties preserved by translation of a language to-and-from the RIF. For example, if it is known that translation from/to rule-platforms R-1 and R-2 to RIF+IFIP+SIM preserves all semantic properties, then a translation from R-1 to R-2 using the RIF+IFIP+SIM as an intermediate step, will result in logically equivalent rule-sets.