Multimedia Systems Project 1

CSE 5348, Fall 2002

Dr. Ramesh Yerraballi

Due Date: 10/6 Midnight

Project Statement

To enhance the performance of the Wumpus world game used under the NSF WISE project, by modifying the current visual percept exchange of uncompressed images to, an exchange of compressed images based on the Adaptive Huffman compression mechanism.
 

Description

Background:
The WISE project at CSE@UTA is a NSF funded Educational Infrastructure project. The official website for the project is http://cygnus.uta.edu/wise/index.html. The project involves the Wumpus world game which is a grid game that is used in AI classes to study various AI search and path planning techniques. In a nutshell, the game involves a grid with obstacles, pits, gold and the wumpus (the enemy), the objective of the player is to avoid the obstacles, the pits and getting zapped by the wumpus and, collect as much gold before exiting the grid. The game has been implemented as a client server program with the client side implementing the player and the server side maintaining the state of the grid and sending percepts (hints) to the client. The server is actually more complex, however for the purpose of this project, this description will suffice.
There are several percepts that the server currently sends to a client, one of which is a visual percept. The visual percept is a ppm image file. The ppm (Portable PixMap) image file format is the simplest form of storing a color image with raw pixel values. The details of the file format can be found by following the link. In the current implementation of WISE the following images are used as a visual percepts:
Pit:     Obstacle:    Wumpus:    Gold:
Each of these images is 32x32 pixels in dimensions with a maximum pixel value of 255. The implementors of WISE foresee several more visual percepts being transmitted in future, some may be real images. These images are currently sent by the server to the client as ppm files, further the actual image that is sent is 64x64 pixels wide with three of the four 32x32 blocks blank (zeroed) and the fourth containing the actual image. Wumpus appears in the top left block, Gold in the top right block, Pit in the bottom left and Obstacle in the bottom right.

Your Project:
Is to change this to apply compression to this file and send the compressed stream to the client. The compression algorithm you are expected to use is the Adaptive Huffman (Lecture 7). You can find some good information to implement the Adaptive Huffman algorithm on the Squeeze page (http://www.cs.sfu.ca/cs/CC/365/li/squeeze/). The page contains src code as well, which you are allowed to use. Note, however that the code has to be understood and modified for it to work for you. The main problem with using this code as is, is with the communication of bits (0s and 1s) which in the applet are not an issue because the applet only simulates bit transfer and does not actually do the physical communication. In your code you will have to address this shortcoming. In order to aid you in this process here is another bit of help. Laurence Vanhelsuwé published and article on "Extending Java Streams to support Bit Streams", in Dr. Dobbs Journal, this is precisely what you need to get the bit transfer support working. In fact, the article describes the use of Bit Streams in the context of another interesting (but not relevant to this project) compression mechanism called LZW.

Now for the nitty-gritty. The code for the Wumpus world is typical of most student written code in that it is very minimally documented, therefore I have made every attempt to minimize the need for you to actually go through the code in great detail. Accordingly I will summarize the code for you. The code is organized as three separate pieces. The Server code (called wumpus code) , the Client code and the Avatar code (which you will ignore for the current project). The Server code by far the most complex, however the crux of the logic relevant to you is essentially contained in the two files - WWServer.java, ImageTransmitter.java. Similarly, the crux of the client code is contained in three files namely, G_Client.java and Frame2.java and ImageReceiver.java. You are expected to ONLY modify ImageReceiver and ImageTransmitter classes. You are welcome to change other files for your debugging needs, but I only expect you to submit your ImageReceiver.java and ImageTransmitter.java in your project submission.

Code Howto:
The code is available to you as two zip files wise_server2.0.zip and wise_agent2.0.zip. Download and unzip them into some working directory for your project. The two zip files must automatically extract to two separate directories (wumpus and client respectively).

Open your favorite command interface (DOS command prompt, Cygwin etc.) in two windows. To compile the server code use javac *.java in one; To compile the client code javac G_Client.java in the other. The client will have to be run first as java G_Client followed by the server as java WWServer.

Configure and run the Server by  following these steps:

  1. Choose "File" option from the menu and pick "Load wumpus world", this will prompt you for a input file. Here is a minimal sample file (layout.def). This should show the world populated with gold, wumpi, pits and obstacles.
  2. This is optional - Choose the "Simulator" option from the menu and pick "Set Configuration" and enter a larger number (say 4) in the "Time to accept connections (in min):" field. The default value of 1 min may be too short for you to start the client.
  3. Now run the server by clicking on the "Start" button that is next to the "Statistics" button. The Server takes a while to startup but you will notice when its ready, from the message in the command window  that reads "To recalculate the wumpus position".
Configure and run the Client by following these steps:
  1. Choose "Plan" from the menu and pick "Display Layout". This will bring up a layout window. Select a starting position by clicking on one of the cells (white). This will prompt you for a direction within the cell to orient the client, choose one and click Close. Now, you can close the layout window to reduce the clutter of windows.
  2. You are ready to run the client, now click on the "Connect" button (Make sure the Server has finished its startup). You will see a dialog message that reads "Request Accepted Successfully". click OK.
  3. You are now connected. Choose "Percept" from the menu and pick "Display Percepts", this will bring up your console for playing the game. The relevant options on this console are under "Action To Be Taken". Pick a move and click action to execute it. You can view on the Server's layout window the result of your move. Also, you will see a little window pop up that gives you the visual percept. You may want to enlarge this for better visibility.
  4. The small popup window will be updated with every move you make. When there is no other entity in your vicinity you will see a black square. When something like a pit, obstacle, wumpus or gold are in front of you, you should see it in the popup window.
That's the game. You should be able to run the code as given to you. However, note that the image is not being compressed. Its your job to compress it.
 

Submission

Instructions will be posted on the class website