2. XML Web Services for Invisible Computing Johannes Helander Researcher Microsoft Research

3. Outline The goals of Invisible Computing Why Web Services? Our approach Table driven serialization Distributed real-time Trust and secure discovery Componentized RTOS Real-time C# Developing code for small devices Educational & research opportunities Availability

4. Why Invisible Computing? The computers stay out of sight and do their job. No setup hassles Make everyday objects better by adding computation and communication Natural user interface – not screen and mouse Rudimentary autonomous operation – added value from services Incremental deployment Devices communicate with each other Devices communicate with big computers as needed

5. Sample Applications Home appliances, security, lighting Medical electronic devices Wearable Computers Robotics, Industrial Control, National Infrastructure Sensor networks Wireless communication gadgets Audio Net Disaggregated PC, smart I/O cards Toys

6. Hardware trends 32 bit microcontrollers are as cheap and power efficient as 8 bit MCUs Single chip computer is a reality Cost close to $5 (“Home depot” price point) No need to aim at lowest point  sweet spot Aggregate of medium volume market is huge  Partially reconfigurable hardware Make hardware easy for software people

7. An Invisible Computing Scenario Interoperability Security Data analysis Power Bandwidth Processing Routing Security Real-Time Non-graphical UI Zero-configuration [VCR] [Pacemaker] XP Embedded [watch]

8. What are Web Services? The general-purpose solution to communication, in XML Convergence of EDI, RPC, MSMQ, app specific protocols and formats Agnostic to underlying transport All about interoperation. Allows partial understanding Across-the-board presentation layer Common protocols obviate need for proxies Builds on critical mass and momentum

9. Do they Scale? XML Web Services conceived to solve e-business interop problem Implementations geared towards high-end computers The same interop problem is the crux of Ubiquitous computing Critical mass required in any business Resource constraints: Silicon – footprint Energy – parsing overhead Bandwidth – size of messages  Efficient implementation and compression

10. SOAP example "Add" request, from PC to NTU simulator, via HTTP then forward to EB63 via encrypted UDP http://schemas.xmlsoap.org/soap/envelope/ <soap:Header soap:encodingStyle=http://schemas.xmlsoap.org/soap/encoding/ >http://schemas.xmlsoap.org/soap/encoding/ http://schemas.xmlsoap.org/ws/2002/05/routing http://172.31.46.26/COB/calc.cob x-udp-aes- soap://172.31.41.244/COB/calc.cob x-udp-aes- soap://172.31.41.244/COB/calc.cob <soap:Body soap:encodingStyle=http://schemas.xmlsoap.org/soap/encoding/ >http://schemas.xmlsoap.org/soap/encoding/ http://tempuri.org/Calc/message/ 14 28 The calculator is a popular interop test

11. Yes, it Works! Implementation shows you can successfully: Realize web services on small low-cost devices, providing good interoperability with PCs and other devices Achieve a high level of security and privacy on those devices Integrate security, discovery, and functional assignment into a hassle-free user experience Setup your home completely independently, yet securely federate with external entities such as e-business Use web services for real-time tasks  Demoed at booth #31

12. Microsoft Invisible Computing A software platform for low cost embedded systems that communicate with each other and with big computers Flexible development for multiple platforms Interoperation with small and big computers Web services and.NET Security and privacy Real-Time Energy aware Low parts cost (targeted for <= $5 computer) Sweet spot: enough for real use and critical mass but no frills XML Web Services: interoperability, tuned for performance Component Based RTOS Standard protocols: TCP/IP, SOAP, PKCS#1, etc..NET virtual machine for C# games or other extensions

13. Invisible continued Interoperates with ASP+ and SOAP Toolkit on Windows XP Client and server, P2P Complete TCP/IP, HTTP, SOAP, Automation, discovery, trust & security, RTOS (dynamic memory, threads, etc), drivers, application with complex data.  Runs in computer with 32KB of RAM, 256KB of ROM. Fewer components  smaller footprint. TCP/IP is biggest hog. Crypto not optimized for size.

14. Outline The goals of Invisible Computing Why Web Services? Our approach Table driven serialization Distributed real-time Trust and secure discovery Componentized RTOS Real-time C# Developing code for small devices Educational & research opportunities Availability

15. Table Driven Serialization Processes messages automatically according to description XML metadata description Compiled offline into compact description Extensible at runtime Process while receiving Zero copy networking Serializer & parser share buffers with network stack & crypto COM-Lite automation Turns messages into object calls Multiple methods in one message Multiple transports and encodings UDP, HTTP, Encryption, Compression Routing, roles, and conversion

16. Distributed Real-Time Experiment in distributed scheduling Real-time data-flow Instigator Producer Consumer Sensor readings Scheduling Sampling

17. Real-Time continued Serialize scheduling trees into XML Reservations pre-declare future activity at each node Instigator of activity orchestrates and tunes reservations based on feedback samples Worker nodes accept/reject schedules  Merge of trees. Location independent. Could write scheduler in XSL. Coordinated schedules allow shared resource scheduling. Could turn off radio. Statistical decision making Confidence test, quality control sampling schedules, probability based admission control Concept demo shown at booth #31

18. Real-Time continued http://tempuri.org/X-Reservation <rs:reservation name=“producer” deadLine=“2004-12-31T00:00:00.5Z“ <rs:reservation name=“producer” deadLine=“2004-12-31T00:00:00.5Z“ tolerance="P456S“ duration="P0.1S"> tolerance="P456S“ duration="P0.1S"> 2000 2000 77 77 <rs:reservation name=“consumer" deadLine="2004-12-31T00:00:00.2Z“ <rs:reservation name=“consumer" deadLine="2004-12-31T00:00:00.2Z“ tolerance="P82S" duration="P0.1S"> tolerance="P82S" duration="P0.1S"> 100 100 </rs:reservation></rs:task> Serialized reservation example Triggers, sub-reservations, resource estimates, tolerances

19. A Secure Invisible Home

20. Setting up a Secure Home Create house authority, e.g. usbkey Touch each device once with usbkey Admits device into trust domain Determines functional relationships heuristically Discovery process finds device with desired function + does key exchange House authority can be offline RSA + AES Write hash of house authority’s key on check to establish trust with bank  Federation of independent trust domains

21. Trust and Discovery Simple SOAP based trust and service discovery for ad hoc networks Integrate trust and functional setup Integrate key exchange with discovery Simple user interaction No external CA required Use Global XML Architecture when infrastructure present Optimized for cluster of nodes. Base station (PC) deals with global issues PKI works on small devices (but can be boosted) 13s RSA decrypt, 0.03s AES on 25MHz Arm7 FPGA takes times down by factors of 3000 and 10000 (3ms & 2µs) Strong crypto necessary for marketability Would people buy surveillance equipment against themselves?

22. RTOS Architecture Support for web services on a chip   General purpose in the abstract. Code and interface reuse.   Special in the concrete. Only take what you need. Component Based Objects everywhere COM interfaces Unified namespace Same interfaces implemented by many components Multiple implementations of any component Specialized to task Pay as you go Late binding and mutation Adaptive to changing requirements Real-time scheduling with application feedback XML based configuration and communication

23. RTOS continued Hardware platforms ARM (many), i386, H8, MIPS, TriMedia, Map1000, 68k, eCOG1 Numerous development boards. Prototype gadgets. Smart I/O cards Can be compiled with numerous compilers ROM sizes e.g. 10KB, 20KB, 200KB on ARM; 26KB, 240KB on x86 Power e.g. 40mW on 5x7 cm 2.8V ARM board with LCD when playing a simple game (snake)

24. It Still Has to be Small! WinXp Invisible

25. Real-Time C# CLR desirable option for embedded systems Great for extensions, games, apps Not practical as the exclusive solution in embedded systems Our real-time scheduling extensions Prototype API implemented Work Item Scheduler allows mixing native and managed threads Native execution stacks are multiplexed

26. Outline The goals of Invisible Computing Why Web Services? Our approach Table driven serialization Distributed real-time Trust and secure discovery Componentized RTOS Real-time C# Developing code for small devices Educational & research opportunities Availability

27. Developing Code for embedded systems using Microsoft Invisible Computing Start with emulation, then simulation, and finally real hardware Debugging on real embedded h/w painful  minimize time spent on this All MS Invisible Computing environments have the same interfaces and basic configurations 1. Winbig 2. NTU 3. Giano 4. Boards

28. 1 – Winbig Runs on Windows XP Uses XP sockets, threads, files i386 binaries Pleasant development under Visual Studio Smallest SOAP stack for Windows XP “big” is the configuration where everything is linked together  usually used for ROM images

29. 2 – NTU Runs on Windows with i386 binaries Implements its own threads and scheduling, etc. Closer to real thing One thread for “CPU”, one for “timer chip” Enables debugging network stack and scheduler under Visual Studio

30. 3 – Giano Hardware simulator Interprets ARM instruction set FPGA simulation enables hardware- software co-design work Easy to add new “hardware” peripherals 14 MHz eb63 board on fast PC Easier to work with than real boards Extremely close to real hardware, except for real-world interactions (e.g. no A/D pins)

31. 4 – Boards Real boards test actual hardware – reality check Development boards still not exactly the same as a real product  Another step closer Instrumentation and monitoring through FPGA co-board JTAG debugging, still unpleasant Most software development done in simulators – very little left to do here

32. Education and Research Microsoft Invisible Computing is a research prototype Experiments in seamless computing through embedded web services Has been used by academia Steve Liu at Texas A&M Open invitation to participate

33. Availability http://research.microsoft.com/invisible Community Source License allows research and education use with few strings attached New code will be added periodically No support available at this time The work presented in this talk was contributed by the MSR Invisible Computing Group Alessandro Forin, Johannes Helander, Behnam Neekzad, Stefan Sigurdsson Special thanks: Paul Pham, Yong Xiong