Proxim-CBR: A Scalable Grid Service Network for Mobile Decision Support

1. Proxim-CBR: A Scalable Grid Service Network for Mobile Decision Support Max Ong Muhammad Alkarouri Rolls-Royce University Technology Centre in Control & Systems Engineering Department of Automatic Control & Systems Engineering The University of Sheffield, UK.

2. Motivation for DAME • Rolls-Royce Power-by-the-Hour • Increasing amount of engine data being collected • New engine monitoring units record up to 1Gbyte of data per flight • Rolls-Royce have over 54,000 engines in service • Total operations of ~10M flying hours p/month • In future, Terabytes of data could be transmitted every day for analysis • Key objectives • Reduce delays • Reduce cost of ownership for the aircraft

3. Grid Computing • The Grid is a scalable network of supercomputers linked via high bandwidth networks • The Grid provides on-demand compute resources for • compute-intensive processes • data intensive applications • collaborative computing • Enables flexible, secure, co-ordinated resource sharing across geographical and commercial boundaries (Virtual Organisations)

4. Introduction to DAME • UK e-Science Grid pilot project (2002-2005) • Funded by EPSRC • To develop and promote understanding of • Grid middleware • Grid application/services integration • Real time issues in Grid computing • Security and dependability issues • Delivered a “Proof of Concept” Grid demonstrator application for the Rolls-Royce aero-engine diagnostic problem

5. Project Partners • Four UK universities: • University of York Dept of Computer Science • University of Sheffield Dept of Automatic Control & Systems Engineering • University of Oxford Dept of Engineering Science • University of Leeds School of Computing and School of Mechanical Engineering • Industrial Partners: • Rolls-Royce plc • Data Systems and Solutions • Cybula Ltd

6. BROADEN Introduction • Business Resource Optimisation for Aftermarket and Design on Engineering Networks (BROADEN) • funded by DTI • Project aim: To build a RR internal GRID as a proving ground for utilizing Web/Grid service technology to fully exploit available IT resources • Jan 2005 – April 2007

7. Engine Modelling & Simulation Service • Perform engine performance simulation for FDI • Based on a complex Rolls-Royce Trent 500 engine model – very compute intensive! • Deployed as a service on the Grid • Accessible by user via a web browser across the Internet • Grid factories enable parallel execution of multiple engine simulation instances • Secure authentication for user and hosts, SSL data encryption

8. CBR Advisor System • Perform fault analysis, knowledge management and data fusion with Case-Based Reasoning technology • CBR links together knowledge that belong together: • information concerning the fault • response to the fault • effects of those responses • Advises maintenance personnel on appropriate maintenance action • Deployed as a service on the Grid • Accessible by remote user via a web browser across the Internet

9. DAME Grid Portal

10. Grid Computing Environment Decision Support Services Mobile Grid-Based Decision Support • High-performance computing On-Demand • Access to Grid-enabled services from mobile devices can be very effective in business environments • Grid computing empowers a remote, mobile user with resources not normally available while working out on the field: • Access to vast computing power • Use of compute-intensive applications • Access to large data repositories • These can be made available via Web/Grid Services

11. Grid Computing Environment Decision Support Services Distributed CBR for Decision Support • High-performance computing On-Demand • Huge quantities of data to process and store • CBR technology is inherently parallelizable • Service Oriented Architecture enables flexible and scalable use of resources • These can be made available via Web/Grid Services

12. Grid Computing Environment Decision Support Services The Case for Mobile Grid Access • Collaborative Decision Support On-line • An aircraft engine expert is usually regarded as a high value resource. • As such, he/she needs to be highly accessible at various times, to investigate problems under demanding circumstances. • Using the Grid as a collaborative computing platform to have a virtual network of experts, who are usually at different geographical locations. • The ability of an engineer to capture investigative findings on-site and instantly upload these findings to the Grid-based system, using mobile access, greatly increases the accuracy of the CBR system in diagnosing the problem.

13. Limitations on a Typical Device • A small device such as a PDA presents various constraints such as • Limited processing power • Limited memory • Small display area • Small network bandwidth • The complexity of platforms and APIs used in such constraints is limited • These constraints were taken into account in the design of a PDA demonstrator

14. Remote Node host environment CBR Service Node Local Cases CBR Search Broker host environment Remote Node host environment CBR Broker Service Proxy host environment CBR Service Node Local Cases CBR Search Proxy Service Registry of service nodes PDA Mobile Monitor Service Remote Node host environment CBR Service Node Local Cases CBR Search Broker registry AQPs ... ... Proxim-CBR Grid Service Network PROXIM-CBR Grid Service Network

15. Service Nodes PROXIM-CBR Grid Service Network • Proxim Service Node • Typically at point of data capture • Contains a CBR Engine • Hosts a repository of cases • Proxim Broker Service Node • Receives requests from proxy service nodes, or directly from CBR portal • Maintains a registry of service nodes • Distributes query requests to service nodes, and summarizes results • Proxy Service Node • Provides access to handhelds • Maintains a registry of available brokers

16. Remote Node host environment CBR Service Node Local Cases CBR Search Broker host environment Remote Node host environment CBR Broker Service Proxy host environment CBR Service Node Local Cases CBR Search Proxy Service Registry of service nodes PDA Mobile Monitor Service Remote Node host environment CBR Service Node Local Cases CBR Search Broker registry AQPs ... ... Demo: Mobile Distributed Health Monitoring PROXIM for PDA-based remote, distributed condition monitoring

17. Demo: Mobile Distributed Health Monitoring • Event Simulator – injects engine fault information at the remote, distributed nodes to simulate engine problem conditions • Flight HKA-8283: Damaged fan blade or mass loss (lock tab) - location Hong Kong • Flight GBA-1313: Single blade loss or oil in drum - location UK

18. Demo: Mobile Distributed Health Monitoring Proxim Mobile Monitor – online, real-time summary from the distributed health monitoring system The problem Confidence of the match Where it occured Operator ID for further online investigation Summary is continuously generated

19. Demo: Mobile Distributed Health Monitoring

20. Example: Remote Ship Condition Monitoring • Event Simulator – injects ship fault information at the remote, distributed nodes to simulate problem conditions • Ship HKISC-8283: - Waterjet steering actuator worn or engine overheat - location Hong Kong • Ship GBSC-1313: Fire in cargo bay or Supervisory Control switched to backup - location UK

21. Conclusions • Grid computing from mobile devices can be very effective in a decision support environment • Service oriented architecture enables business/engineering processes to be examined so that services can be provided to 'accurately and sufficiently' support core operations, providing flexibility for a more dynamic management of resources and processes

22. Recent Papers • “Service-Oriented Architecture on the Grid for Integrated Fault Diagnostics” • Journal of Concurrency and Computation • “Scaleable Grid Service Network for Mobile Decision Support” • E-Science AHM2006 conference (Sept 2006) • “Grid-Based Decision Support with Pro-Active Mobile Computing” • IEEE Int'l Conference on Services Computing (IEEE-SCC2005) • “Decision Support System on The Grid” (Best Paper Award) • Int'l Journal of Knowledge-Based & Intelligent Engineering Systems • “Modelling, Optimisation and Decision Support Using the Grid” • E-Science AHM2005 conference