SILF

School Internet Laboratory of Physics

Jacek Bzdak


Silf is a remote laboratory,
but also a framework for your
remote laboratory.









Example experiment

Promieniowanie Gamma

  • When gamma radiation is high-energy photons created by nuclear reactions
  • When passing through matter gamma radiation is attenuated
  • This attenuation is governed by differential equation: 
 dI=mdx
  • Probability of absorbtion of quanta for each part of path is dependent only on material type
  • After integration we get: 
 I=Io*exp(mx)

Example results

Results in SILF

LIVE EXAMPLE

Introduction


  • This is our second Remote Lab
  • We had insights on what we can do better

ILF


  • Our old project
  • Still working!
  • Client is based on Java software
  • Java was then only way to run it on multiple platforms

ILF Architecture 

SILF ARCHITECTURE

Another LOOK

Client

Client software


Java disadvantages


  • Java must be installed
  • Now is unsecure and it's instalation is discouraged on client machines
  • Client weights about 3MB (could be smaller)
  • Despite being stable, we had bunch of backwards incompatible changes in Java
  • Each experiment has it's own client
    • Despite plugabbility it still takes time
    • Not only programming but deployment!

NEW CLIENT


  • Use HTML5 
    • also AngularJS and Coffeescript
  • Better yet: have single client for all experiments

Metadata driven Lab


  • Experiment sends GUI description to users
  • GUI is built on the fly 

Metadata DRIVEN LAB


Advantages:

  • There is one single client
  • Existing experiments "get features" as we develop the client.
  • Easy (for example) to build Android client.


Metadata DRIVEN LAB


(more on this when we discuss protocol)

SERVER

SILF ARCHITECTURE

SerVER


  • We use stock XMPP server
  • With some minor adjustments
  • It's fast (way faster than we'll ever need)
  • No work for us! 


Reservation enforcement


  • Reservations are enforced using XMPP moderation tools
  • Each experiment has associated chatroom 
  • By default users can only read messages from chatroom
  • Experiments always can send messages
  • During reservation user temporarily gets voice
  • This is implemented by the athena bot

Athena bot


  • Takes part in enforcing reservations
  • Interfaces with WWW via the REST API 
  • Creates and manages experiment XMPP rooms
  • Will store experiment results for later retrival
  • Will enable observer users
  • Written in Python


WWW Page


  • Manages existing experiments
  • Manages reservations
  • Written in Python/Django

Experiment


  • Experiment is a normal XMPP client
  • After connecting it joins a particular XMPP room
  • Everything that speaks our dialect of XMPP is an experiment
  • Protocol is documented!



Experiment


  • Experiment is concerned only with
    performing the experiment.
  • It is not aware of:
    • chatrooms
    • reservations
    • data storage
    • who is using the experiment
      (user or a bot)

Our implementation of experiment


  • Every device has the same API
  • Every device is isolated in it's own process
    • This enables us to wotrk with third-party drivers that are less-than stable.
  • Operations on device are asynchroneous for speed
    • Sometimes drivers require us to wait
    • This may be changed --- as asynchroneous code is hard  to write

SILF BACKEND COMMONS

Protocol


  • A dialect of XMPP
  • 100% XMPP Compiliant!
  • Uses XMPP groupchat messages
  • Each of them contains a <labdata>
    tag

Labdata


Labdata tag:

  • Namespace defines action
  • type
    • query
    • done
    • result
    • error
  • Content depending on type
    and namespace

LABDATA CONTENT


  • Labdata content is in JSON language
    • It is very flexible, has basic types
    • Also parsing of json is free in browser,
      and easy everywhere

Example conversation


Example Conversation

EXPERIMENT METADATA

{
    "experimentId": "urn:uuid:a12e6562-5feb-4046-b98f-d8c40ca1609c",
    "mode": "aluminium",
    "settings": {
        "acquisition_time": {
            "live": false, "metadata": {
            "label": "Czas pomiaru dla jednej grubo\u015bci materia\u0142u"},
            "name": "acquisition_time", "validations": {"min_value": 5.0, "max_value": 600.0}, "type": "interval",
            "sort" : "1"},
        "light": {"live": false, "metadata": {"label": "O\u015bwietlenie", "style": "light_control"}, "name": "light", "type": "boolean", "sort" : "1"}
    },
    "resultDescription": {
        "time_left": {"name": ["tile_left"], "type": "array", "class": "interval-indicator", "settings": {}, "metadata": {"label": "Czas pozosta\u0142y do ko\u0144ca bierz\u0105cego punktu"}},
        "chart": {"name": ["chart"], "type": "array", "class": "chart", "settings": {}, "metadata": {"chart.axis.y.label": "Liczba zlicze\u0144 zarejestrowana przez licznik", "chart.axis.x.label": "Grubo\u015b\u0107 przes\u0142ony [mm]"}
   }
}

Example settings


<labdata xmlns="silf:series:start" type="query">
{
    "acquisition_time": {"value": 150, "current": false},
    "light": {"value": false, "current": false}
}
</labdata>


Experiment bot


  • Bot that performs experiment
    is quite easy to create
  • It just needs to send proper messages
    to an XMPP room
  • Or maybe do some data analysis on
    to highlight interesting parts



REMOTE Experiments


  • Integrating remote experiments is easy
    • If these experiments would not need to change our client
    • Very minor adjustments to out stack would be required
  • Remote experiment would just need to connect to our XMPP server
  • We'd just need to configure chatroom for it.

Conclusion

Successes


  • Software developement costs per experiment
    are very low (really negligible);
  • Existing infrastructure gets "free lunches"
    --- exisiting experiments are enchanced for free;

Challenges


  • Keeping experiments running
    requires either large up-front costs and/or
    constant maintenance.
    • We try to increase reliablilty.
    • eg. whenever there is error in WWW page,
      or on my experiment I get an e-mail.
  • Changing SILF to be deployed on a
    nationwide scale would require
    mass-production of electronics. 

Open-source


E-Textbook


  • Source files are in DITA format
  • DITA is XML based
  • DITA stands for Darwin Information
    Typing Architecture
  • DITA allows for easy content
    re-arrangement

E-TEXTBOOK


  • DITA Open Toolkit is used to generate
    XML and PDF output formats
  • You can see HTML version here.

Opportunities For cooperation

Opportunities


  • You can "just" use our open-source software
    • However it would be easier to have consultants who created the Infrastructure
  • We can provide working remote laboratory experiments
    • Infrastructure could be provided by us
    • or by You
  • We could cooperate in creating similar infrastructure in your institute
  • We could create multimedia content

Multimedia

  • Movies
  • Animations/Virtual Experiments
    • In Java FX
    • In Javascript
    • In LabView
    • Flash
  • We have people that worked on CERN Master-Class programme


Thank you for attention

Made with Slides.com