CS422

Lab 3: Implementing an Internet Telephone

Slides

PSO Notes on Voice Capture

Here is the Grading form. This grading form is very similar to the one that will be used during the presentation. However, the one used during the presentation could be slightly different. Also, notice that 20/100 points are reserved for code inspection.

Goal:

In this lab you will implement an Internet Telephone. You will use Windows Microsoft Visual Studio .NET and Windows. You may work on any of the ITAP PC labs that already have Visual Studio.This project will be done in teams of up to four people. You may decide to work individually or in teams of two members but the same grading criteria will be applied to everybody.
 

Recommended Reading:

Getting Started with Visual C++
Windows Sockets : How to use sockets in windows.
Download DirectX SDK. This contains the libraries and documentation you need for capture and playing sound.

DirectX/DirectSound: Capturing and playing sounds in Windows

msdn.microsoft.com: Other documentation you will need.

Download Visual Studio Free for Purdue Students

Architecture:

The Internet phone system will consist of two programs: The telephone program will have the following subsystems: The protocol used by the Directory Server/Client, Ringer Server/Client will be TCP to make the lookup and call start/end reliable. The Voice Capture Client and the Voice Player Server will use UDP since the voice samples should arrive on time to be played on time and using TCP/IP may add cumulative delays in the voice due to the retransmission.
 

Step 1: Download and unzip the example-code

Download and unzip the file Internet-Phone.zip. This file contains three examples:
  1. UDPTimeServer  is a UDP time server that shows you how to use UDP sokets in Windows.
  2. UDPTimeClient:  is a UDP time client that shows you how to use UDP sokets in Windows.
  3. Capture-Play is a program that shows you how to capture and play sounds in Windows using DirectX/DirectSound
Unzip this file and run your Microsoft Visual Studio .NET. Open the UDPTimeServer, build it and run it. Also Run another instance of Microsoft Visual Studio .NET. and open UDPTimeClient, and build (do not run it). These are console applications, that means that you can only run them from a command prompt as they do not have a GUI. The executable files will be in the Debug directory. To run the client, first run a command prompt ( Start->Run and type "cmd") and then cd into the directory UDPTImeClient/Debug. Then type "UDPTimeClient localhost". You may try these programs in different PCs.
 

Step 2 Creating the Basic GUI

Using the wizard in Visual Studio implement the initial GUI for the phone program. Select New->Project->MFC-APplication. In the application name field type "Internet Phone" and in the Location select the directory where you unzipped the files in step 1. In the Application type select Dialog Base then select Finish.. That will create a new application. Try playing around with Visual Studio to be familiar with it.

The initial GUI will be dialog based and it will have the following components:

Step 3 Implementing the Directory Server and Client

Implement the Directory Server. Create a Dialog-Based application that will have the list of current users and the IP addresses, and a button to exit. The server will keep in memory the list of pairs <name, IP-address> that will be added by the Directory Clients. Use TCP sockets for this server. The server will be waiting for incoming connections and when a request of the following form arrives:

ADD<crlf>
<Name><crlf>
<crlf>

it will register this entry in the Directory Server. The server will return:

OK<crlf>
<crlf>
 or

ERROR<crlf>
<Error message><cr><lf>
<crlf>

 The IP address will be taken from the socket information returned by accept.

When a request of the form:

DELETE<crlf>
<crlf>

is received, the entry that corresponds to that IP address will be removed.

The server will return:

OK<crlf>
<crlf>
 or

ERROR<crlf>
<Error message><cr><lf>
<crlf>

To get all the user entries the client will send:

GET<crlf>
<crlf>

and the server will answer with the list of user/ip pairs
User1<crlf>
ip address 1<crlf>
User2<crlf>
ip address 2<crlf>
...
<crlf>

The end of the list will be delimited by an empty line. Alernatively, the server will reply:

ERROR<crlf>
<Error message><cr><lf>
<crlf>

The telephone program will send an ADD message when the telephone program starts.
It will also send a DELETE message when the telephone program exits.

Also, add the Directory client to the GUI from Step 2.

Turnin step 1-3 before March 9th 11:59pm. Zip all the files and then copy the files
to one of the lab machines (Use sftp for example). Then type the turnin command: 

turnin -c cs422 -p lab3-1 Internet-Phone.zip

Step 4 Implementing the Ringer Server

Note: Try to finish this step before the deadline to continue with step 5. Step 5 may take more time to get it right.

Implement the Ringer Server and Ringer Client. The Ringer Server will be started when the phone program starts.

When the other side sends the request:

RING<crlf>
<name><crlf>
<call-number><crlf>
<Sampling-speed><crlf>
<crlf>

The server will replay with the following when the user clicks the incoming-call button:

PICKUP<crlf>
<name><crlf>
<call-number><crlf>
<Sampling-speed><crlf>
<crlf>

The <name> is the name of the user. <call-number> is the number of calls the user has made/received since the program started. The sampling speed used for capturing voice.

or if there is a call in progress:

BUSY<crlf>
<crlf>

or if there is an error:

ERROR<crlf>
<Error message><cr><lf>
<crlf>

The connection will be kept open until one of the parties decides to hang up. If one of the parties decides to hangup it will send the message:

HANGUP<crlf>
<crlf>

and then both close the connection.

Integrate the Ringer Server and Client in the phone program. The Ringer Server can be started in a different thread that is started when the program starts.

Turnin step 4 before March 30th before 11:59pm. Zip all the files and then copy the files
to one of the lab machines (Use sftp for example). Then type the turnin command: 

turnin -c cs422 -p lab3-2 Internet-Phone.zip

Step 5 Implementing the Voice Capture Client and the Voice Player Server

To understand capture and playback, read first DirectX/DirectSound. Then, take a look at program Capture-Play given to you that plays what is captured in the microphone. I recommend you to buy a earphone/microphone headset. You can get a good quality one for less than 20 dls.

Build and run the Capture-Play program from Visual Studio. Then select the capture and playing speed. You will hear what you speak with a small delay.

The heart of this program is a thread that reads from the capture buffer and writes to the play buffer.You will need to separate this thread into three. For the Voice Capture Client, when the call is started create a thread that will wake up every time there is new data captured and it will send it to the other peer in a UDP datagram. The Voice Player server will have two threads: one thread receiving the UDP packets and copying the voice to an intermediate data buffer (called jitter buffer) and another thread that will wait until the play buffer needs more data and it will copy the data  from the intermediate buffer. At the beginning fill the jitter buffer with the contents of two packets before starting playing.

The format of the UDP packet will be the following: 

unsigned long callNumber;       // Call number from when the call started.
unsigned long sampleOffset;     // Position in the stream of the first sample in the packet
unsigned long numberOfSamples;  // Number of samples in the packet

unsigned short samples[700]     // 700 samples of voice max.
The callNumber will be the call number used when the call started. The sampleOffset is the position of the first sample in the packet in the whole stream. sampleOffset is used to play the sample on time. The numberOfSamples will be 700 most of the time except maybe at the end.

Use 8000Hz Mono for the capturing frequency. This frequency will generate about 12 UDP packets per second. You may try experimenting with other speeds.  Also add  in the dialog of the phone program the number of seconds the call has used and the number of packets lost/delayed.

Step 6 Implementing Conference Calling

You will also implement a conference calling feature. When a call is already in progress where A called B, another caller C will be able to join the call. C will dial  B's number, and instead of C getting a busy signal, a dialog will be opened on B asking if C can join the conversation. C will get a "dialing signal" while all this happens. If  B allows C to join in, C will be able to talk/listen to both A and B and vice versa, otherwise C will get a busy signal. The number of  participants in the conference will be unlimited.

All callers in a conference call will have to call the host's number. If another host D calls a caller like A or C, D will get always a busy signal. A conference call will end when the host A hangs up or when there are no more callers in the call.

Step 7 Other files you need to submit

Along with your step 5 and step 6, you also need to turn in the following:

Create a README file with

a) The name of your teamates.
b) What required features in your project did not work.
c) List of extra features.


Also  create a file named "TEAMMATES" with the login and names for each team member. The login and the name should be separated by a colon and one team member per line.

Example:

jsmith: John Smith
psellers: Peter Sellers
dmessing:Debra Messing
gclooney: George Clooney


Final Turnin

Turnin step 5, 6 and 7 before  April 5th before 11:59pm. Zip all the files and then copy the files
to one of the lab machines (Use sftp for example). Then type the turnin command: 
turnin -c cs422 -p lab3-3 Internet-Phone.zip
You will present your project during  the PSOs on April 6th. We will post a schedule with the possible slots  before that date.

Other things to do -- extra credit

You may decide to add some of the following features.