IRISNET Documentation

Table of Contents

• Installation
• Software Dependencies
• Sensing Agents
• Webcam
• Organizing Agents
• Logger
  
• Technologies, API's, and Standards
• Coding Style and Documentation
• Test Suite
• Parking Space Finder Application
  
• Developing a new application[ps][pdf]
  

Installation

The easiest way to install IRISNET is to check out everything from the CVS repository and do a full installation.

• Download
  • Follow instructions on.  http://iris.wv.cc.cmu.edu/download.html  After download, run "install" in the main IRISNET directory.
• Set environment variables
  • Add "IRISNET/packages/opencv-0.9.3/otherlibs/highgui/.libs/" and
    "IRISNET/packages/opencv-0.9.3/cv/src/.libs/" to your LD_LIBRARY_PATH environment variable.  These environment variables tell where OpenCV is installed.
  • If Java is not installed on your system, add "IRISNET/packages/jdk1.3.1_06/jre/bin" to your path.
• Configure webcam
  • You need to install the appropriate webcam drivers for your webcam.  For USB Quickcam 3000 Pro webcams, use the driver on http://www.smcc.demon.nl/webcam/
  • As root, run "insmod -f pwcx-i386.o" to load the Quickcam drivers.
• Run Test Suite (optional)
  • You can optionally run "testall.sh" in the IRISNET/TestSuite directory to see if everything is installed correctly.

Software Dependencies

Sensing Agents

• Sensing agent coordinates the loading and execution of senselets.
• Dependent on Webcam for getting sensor feed.
  
• Intel OpenCV Library
• Intel's open source computer vision and graphics library.
  
• Intel Site: http://www.intel.com/research/mrl/research/opencv/
• Sourceforge Site: http://sourceforge.net/projects/opencvlibrary/
• More Documentation: http://limu.is.kyushu-u.ac.jp/~yosimoto/other/opencv-docs/
• Intel IPP/IPL
• OpenCV used to depend on Intel's Image Processing Library (IPL).  Therefore, OpenCV still shares a lot of data structures with the IPL.  IPL is being moved into Intel's Integrated Performance Primivites (IPP), which is not free.  OpenCV will use IPP to improve its performance if IPP is available, but is not dependent on it.  IPP is available for both Windows and Linux platforms.
• Intel Site: http://www.intel.com/support/performancetools/libraries/ipp/ia/
• Richard Stevens Unix Network Programming Library
• For a lot of socket calls, we use the unp wrapper library.
• http://www.kohala.com/start/
• CSAPP
• We also some the csapp wrapper for some socket calls.
• http://csapp.cs.cmu.edu/
  

Webcam

• Webcam loads video frames into shared memory for use by the senselets.
• Camstream
• C++ wrapper for Video4Linux and demo application.  We use its libraries for ease of implementation, but in theory we can call the Video4Linux API directly.
• It has a nice GUI that lets us test the cameras easily.
  
• Camstream is dependent on Qt.
  
• http://www.smcc.demon.nl/camstream/
• Qt
• A cross platform C++ GUI toolkit.
• Comes with KDE.
• http://www.trolltech.com/
• USB Quickcam drivers
  • We have a set of USB Quickcam 3000 Pro webcams, which the following web site provides Video4Linux drivers.
    
  • http://www.smcc.demon.nl/webcam/
• Video4Linux
  • API used by Camstream and other programs to access video devices under linux.  Most webcam drivers provide this interface.
    
  • http://www.exploits.org/v4l/
• Shared Memory API
• We use shared memory to pass data from Webcam to the senslets.
• See man pages for shmget(), shmat()
• Quick tutorial at http://www.cs.cf.ac.uk/Dave/C/node27.html
  

Organizing Agents

• Java 1.3.0
• We use 1.3.0 instead of 1.4 because Xindice has not been tested with 1.4.  There have been known problems with Sun Java SDK 1.3.1.
  
• Apache Xindice XML Database: http://xml.apache.org/xindice/
• Apache Xerces2 Java XML Parser: http://xml.apache.org/xerces2-j/index.html
• Apache Xalan Java XSLT Processor: http://xml.apache.org/xalan-j/index.html
• sixpath XPath Parser: http://sourceforge.net/projects/sixpath
  

Logger

• All IRISNet nodes send debugging to the logger, which reconstruct the flow of packets in the network.
  
• Nam
• Network animator used for giving demo's.
• We modified Nam to show different colors and window sizes, etc.
  
• http://www.isi.edu/nsnam/nam/
  

Technologies, API's, and Standards

XML

• We use XML as a distributed database for each application, and also as a data format for network communication.  XML is handled mostly by the Organizing Agents.
• We rely heavily on Apache XML libraries.  http://xml.apache.org/
  

Coding Style and Documentation

• Java:
• Coding style: http://java.sun.com/docs/codeconv/
• Javadoc: http://java.sun.com/j2se/javadoc/
• C/C++:
• "indent -i 8 -kr < ugly.cc > pretty.cc"
• Doxygen: http://www.stack.nl/~dimitri/doxygen/index.html

Test Suite

The test suite tests the components in system.  All tests are in IRISNET/TestSuite/.  Within each directory, there is a script called testall.sh that runs the tests the components.  Also within each directory, there is a testcases.txt file that describes each test in detail, the possible causes of failure for each test, and possible remedies.  Listed below are the tests and the two letter code of each test in parentheses.

• Network (NT)
  • Pings various sites to see if network connectivity is good.
• Sensing A
gent (SA)
• Builds the dummy2 application and tries to have the SA load it.  Checks to see that both are running.
• Video Test (VD)
  • Checks whether there are any compatible video devices on the system.  Checks that we can get a live image from the video device.
• Webcam (WC)
  • Checks that the webcam utility is working correctly.  Runs the webcam and checks that we can get a live image from it.
    
• OA
(OA)
• Checks whether Java is present on the system and that we can start Xindice.
  

Parking Space Finder Application

The parking space finder application is an example application to demonstrate what can be done with IRISNET.  It uses the video feed to detect which parking spaces are full or empty, and reports them to the OA's.  There is a web based GUI that queries the OA's to find parking lots with free spaces.  The application files are in IRISNET/Applications/parking/

• XML Schema for OA
  • Files are located in IRISNET/Applications/parking/OA
  • The example schemas have hard coded IP addresses of our demo machines.
• Web based front end
  • Files are located in IRISNET/Applications/parking/FrontEnd
  • The example html forms have hard coded IP addresses to our demo machines.
• Application code for SA
  • Files are located in IRISNET/Applications/parking/SA
  • These files are general enough to be run on any SA

Running the Parking Space Finder on SA

We now describe the process of setting up an application on the SA.  The Parking Space Finder files can be used as a template to create other applications.

• Compiling and Installing
  • Run "configure" in IRISNET/Applications/.  This will generate the proper make files for SA applications.
  • Run "make" in IRISNET/Applications/.  This will compile the common libraries and the applications.
  • In IRISNET/Applications/parking/SA/, run installSA.sh.  This copies the parking space application to the proper SA directories.
  • Similarly, run "configure" and "make" in IRISNET/SA/IrisWebcam/.  This builds the application to load images into the shared memory.
  • Run "configure" and "make" in IRISNET/SA/IrisSA/ to build the SA
  • Run "make" in IRISNET/Util/SAControl/ and also IRISNET/Util/PortSink/.  These are helper applications that facilitate running and debugging SA applications without an OA running.
• Running the application
  • Open up a new terminal and run IRISNET/Util/PortSink/portsink 6789.  This program listens on port 6789 and reads all data from that socket.  It simulates an OA listening for SA data.
  • Do one of the following:
    • cd into IRISNET/SA/IrisWebcam/.  Run "./loadImage images/calibrate.bmp 75".  This loads the calibration image into shared memory key 75.
    • As root, load the Quickcam driver.  Not as root, run "runwebcam.sh" in IRISNET/SA/IrisWebcam/.  This loads the live sensor feed into the shared memory.
  • cd into IRISNET/SA/IrisSA/src/parking/.  Run "./calibrate > pspots.conf".  This program finds where the parking spots are from the calibration image and writes the reference images for empty parking spots to "pspot?.jpg".
  • run "./updateconf".  This copies the reference images and calibration configuration to the proper SA directory.
  • cd into IRISNET/SA/IrisSA/src/.  Run "./runsa".  This starts the SA, listening on port 3456 for commands from the OA.
  • In IRISNET/Util/SAControl/, run ` ./SAControl 127.0.0.1 3456 "S 127.0.0.1 6789 parking" ', without the single quotes.  This simulates the OA asking the SA to start the parking space finder application.  You should see an image pop up displaying the free and empty parking spaces.
  • cd into IRISNET/SA/IrisWebcam/.  Run "./loadImages".  This script loads a series of images into the shared memory, simulating cars coming into the parking lot.  The parking space finder application should reflect the updated status of the parking lots.
  • Finally, in IRISNET/Util/SAControl/, run ` ./SAControl 127.0.0.1 3456 "U 127.0.0.1 6789 parking" ' to terminate the parking space finder application.