Network-centric Middleware for Service Oriented Architectures across Heterogeneous Embedded Systems

Transcription

1 Network-centric Middleware for Service Oriented Architectures across Andreas Wolff, Jens Schmutzler, IEEE International EDOC conference, Workshop on Middleware for Web-Services, Annapolis, USA, October 2007 University of Dortmund Communication Networks Institute (CNI)

2 Overview Motivation & Requirements for Embedded Web-Services The µsoa Approach MORE Middleware Architecture Deployment Example Conclusions & Outlook Slide 2

3 Exemplary End-user scenario: Environmental monitoring Universität University of Dortmund Processing Level Relay Level Sensor Level Mitigation Manager (Fire brigades, forest department,...) Subscribe information of certain sensors Challenge: Energy, communication link and real time constraints Slide 3

4 Requirements for embedded Web-Services Real-life Scenarios Environmental Monitoring & Mitigation Management Remote Chronic Care Support Common requirements Distributed and connected via Internet using integrated Web-Services Heterogeneity of devices Embedded system conditions Aims of the MORE project: Generic Middleware for resource constraint scenarios Reduce deployment time, Reuse of services Slide 4

5 Resource constraints of Embedded Systems Example Sensor Relay Level: mm mm <10g 2.5g Limited Processing power Limited Memory 45 mm 20 mm Wireless Moteiv s Tmote Module mini TC65 Technical Specifications Supercomputer Limited Energy ARM TI MSP430 v7 (8MHz) 2x 8x ADC Interfaces inputs 2x serial DAC interfaces outputs with the ITU-T V.24 protocol USB Multiple 2.0 interfaces full speedfor UART, SPI, I2C, and 1-wire Memory: KB (RAM) and KB MB (Flash) Wireless Technologies GSM/GPRS 250kbps 2.4GHz IEEE Chipcon Wireless Transceiver (CC2420) Slide 5

6 Measurement of the impact of running simultaneous tasks on a Embedded System (TC65) 1000 Execution Delay nearly doubled 900 Status request Incoming call Execution Delay [ms] One concurrent task I/O Interrupt Ping (5 times) Loss of GSM connection Exemplary Periodic operation Measurement test cycles Slide 6

7 Processing standard SOAP leads to unacceptable delays & energy consumption Embedded System XML Parsing Benchmark Universität University of Dortmund Required parsing time (in ms) s 4000 Exemplary SOAP Message (600 bytes) Platform: Siemens TC65 Wireless Module J2ME kxml 2.0 Pull Parser Target area Number of parsed bytes of XML file High Energy consumption due to long parsing times Reduced real-time capabilities Slide 7

8 Solution Approach: µsoa 598 Bytes function name <?xml version="1.0" encoding="utf-8" standalone="no"?> <SOAP-ENV:Envelope SOAP-ENV:encodingStyle=" xmlns:soap-env=" xmlns:soap-enc=" xmlns:xsi=" xmlns:xsd=" <SOAP-ENV:Body> <ns1:getvalue xmlns:ns1="urn:mysoapservices"> <param1 xsi:type="xsd:int">123</param1> </ns1:getvalue> </SOAP-ENV:Body> </SOAP-ENV:Envelope> Standard SOAP message service name input parameter µsoa Proxy Service Universität University of Dortmund Available implementations 100% Standard SOAP Zipped SOAP SOAP Management Service Binary SOAP µsoa µsoa µsoa µsoa µsoa 2.5% 60% 20% µsoa Intercepting service advertisements for context configuration Embedded Service Node Internode communication Embedded Service Node Embedded Service Node Compression Method µsoa in detail Slide 8

9 MORE Service Oriented Architecture Service Connectors Universität University of Dortmund Internal Sensor Standard SOAP Implementation of Service Functionality (Service Logic) µsoa Inter-Node/Inter-Service Communication Classical Layered View & Service Oriented View Core Management Service Different Service Connectors Slide 9

10 MORE enabled deployment Processing Level Relay Level Sensor Level µsoa Proxy Service Low energy consumption of sensors by efficient handling of measuring / broadcasting intervals Group Management Services End users specify, which sensors are of interest to them (spatially explicit information) Communication Services Communication of and between nodes Measurement Services Sending measured data to the transmitting device / georeferenced information Slide 10

11 Conclusion MORE middleware: development and validation of new concepts to enable Web-Services for Embedded Systems Service-oriented Architecture and exemplary use case scenario of a mitigation management system: Fusion of sensors with Web-Services Reduced deployment time and reuse of services Outlook: Validation of experimental system by real end users In-depth performance evaluation: focus on reliability and scalability Slide 11

12 Thank you for your attention! For further information, please contact: Slide 12

13 Backup: MORE Acronym Network-centric Middleware for Group communication and Resource Sharing across Slide 13

14 MORE Group Management Service Universität University of Dortmund Group A Group C Group B Control of dynamic groups in MORE Policy based Group Management Policies enable a higher reliability and in combination with group communication cost effective efficiency gains Slide 14