Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
AMPol: Adaptive Messaging Policy Raja N. Afandi, Jianqing Zhang, Munawar Hafiz, Carl A. Gunter Computer Science Department, University of Illinois Urbana-Champaign Presentation Date Dec 05, 2006 Adaptive Policy • Large scale systems often have diverse policies – Many administrative domains – Difficult or impossible to impose a uniform policy on all participants (“Global standards make the system brittle” – Alan Karp) • Support for non-functional (QoS) features such as security and reliability breaks the interoperability of the system – Constraints may change frequently Slide 01 of 18 Strategy • Service Oriented Architectures (SOAs) based on Web services offer a promising platform • Basic strategy: responder advertises WS-based policy, initiator adapts dynamically • Basic architecture based on three components – Policy model describes declarative domain-specific policy rules – Policy discovery governs how to publish, find, and merge policies – Extension and Enforcement (EE) provides means to add capabilities and enforce policies Slide 02 of 18 Our Contribution • New Case Study - WSEmail extension with novel features • Dynamic Extension - Policy Extension - System Extension • Multi-node operations - Message Sender, Recipient and Multiple Intermediaries Slide 03 of 18 WSEmail • WSEmail: Internet messaging based on Web services • Uses XML, SOAP, XMLDSIG, WS-Policy, etc. rather than SMTP, IMAP, POP, S/MIME, etc. • Improves security, flexibility, integration Lux May Bhattad Gunter ICWS 05 Slide 04 of 18 AMPol • Adaptive Messaging Policy (AMPol) is an SOA for policy adaptation in messaging systems such as email, instant messaging, etc. • Instantiates three components: Policy model, Policy discovery, Enforcement and Extension • Extends WSEmail with modest changes to underlying system • Illustrates benefits to adaptive policies Slide 05 of 18 Problem Scenario RMTA reality SMTA wonderland To: bob@reality From: alice@wonderland Sender MUA alice@wonderland Recipient MUA bob@reality Slide 06 of 18 Specifying the Policies All messages must be signed SMTA wonderland Egress policy of Wonderland Alice must not sent .bmp extensions Egress policy of Alice RMTA reality Ingress policy of Reality Ingress policy of Bob Alice must use encryption To: bob@reality From: alice@wonderland Sender MUA alice@wonderland Recipient MUA bob@reality Slide 07 of 18 Policy Model • The policy constructs should be distinct, modular and extensible • Static policy and dynamic policy • Supports – Meta-specification for policy enforcement – Rule prioritization to resolve conflicts – Public vs. private rules • Domain specific policy rules: APES – – – – Attachment Payment Encryption Signature Slide 08 of 18 APES • Attachment. Types of attachments allowed in mail messages. Example: No .pif extension file can be attached. • Payment. Type of cost imposed on message senders. Example: A message sender must solve an RTT (Reverse Turing Test) Puzzle. • Encryption. Type of Encryption Mechanism. Example: 3DES encryption is used. • Signature. Type of Signature Mechanism. Example: Messages are signed with SHA-1 Hash. Novel Technologies supported by our policy model * Hashcash * Reverse Turing Test (RTT) * Identity Based Encryption (IBE) Fenton Thomas P.E.T.Mail, Voltage IBE Library Slide 09 of 18 Policy Discovery • Policy Advertisement • Policy Merging • Policy Query (Policy Query Protocol) Slide 10 of 18 Policy Advertisement Alice must sign packet Enforcement Point: wonderland Alice must use IBE Enforcement Point: reality Egress policy of Wonderland SMTA wonderland Alice IBEHashcash Alicemust mustuse solve Enforcement Point: reality Enforcement Point: Bob Merged Static Receiving Policy of reality domain Alice must sign packet Enforcement Point: wonderland Alice must solve Hashcash Merged Static Enforcement Point: Bob Egress policy of Alice Sender MUA Sending Policy of alice@wonderland wonderland domain RMTA reality Ingress policy of Reality Ingress policy of Bob Recipient MUA bob@reality Slide 11 of 18 Policy Merging RMTA reality SMTA wonderland Static sending policy of wonderland domain Static receiving policy of reality domain Dynamic Policy Alice must sign packets Enforcement Point: wonderland Alice must use IBE Enforcement Point: reality Sender MUA alice@wonderland Alice must solve Hashcash Enforcement Point: Bob Recipient MUA bob@reality To: bob@reality From: alice@wonderland Slide 12 of 18 Extension RMTA reality SMTA wonderland Requires Extension for Hashcash Sender MUA alice@wonderland To: bob@reality From: alice@wonderland Recipient MUA bob@reality Hashcash plugin Slide 13 of 18 Enforcement Check that packets are signed Check that packets are encrypted with IBE SMTA wonderland Dynamic Policy RMTA reality Alice must sign packets Enforcement Point: wonderland Alice must use IBE Enforcement Point: reality Alice must solve Hashcash Enforcement Point: Bob Sender MUA alice@wonderland Finally! --- Sincerely, Alice Recipient MUA bob@reality Check the Hashcash puzzle solution Slide 14 of 18 Enforcement and Extension • Policy Framework – Extracts the policy conformance and enforcement logic into independent and dynamically pluggable extensions – Core policy engine is generic enough to process many complex constraints – Unlike WS-Policy framework in which assertions are domain specific and any new addition requires an update to core policy engine • Extension Framework – Manages extensions – Have policies to control the download and execution of extensions – Download plug-ins from secure third-party plug-in server Slide 15 of 18 EE Sub-components Slide 16 of 18 Summary • End-to-end solution for supporting non-functional (QoS) policies • Reference architecture for adaptive middleware for messaging systems • Application of this middleware for email system based on Web services • Validation of proposed approach through a case study • Addresses gaps in prior work: – Multi-node operation – Dynamic extension Slide 17 of 18 Current and Future Work 1. 2. 3. 4. 5. Semantic Web QoS (AMPol-Q) Policy Merging (Based on Defeasible Logic) Attribute Based Messaging (ABM) WS-Email for Health Alerts Learn More: http://seclab.cs.uiuc.edu/ampol Slide 18 of 18 Backup Slides AMPol Policy Model Backup Slide 01 WSEmail Case Study Backup Slide 02