PHONE BASED E-MAIL SERVER
Professor Tzi-cker Chiueh,
Experimental Computer Systems lab,
State University of New York at Stony Brook
Ph.D., University of California
Adil Attari,
Experimental Computer Systems lab,
State University of New York at Stony Brook,
M.S, State University of New York at Stony Brook
Abstract :
As electronic mail is becoming a standard way of communication, the ability to send e-mail even when you do not have physical access to a machine hooked to the Internet is of tremendous convenience. Using a touch-tone phone to dial-in to a machine
that will carry on your e-mail delivery seems to be a good solution for many situations.
Today in the industry we find products that exploit this idea, yet requiring an additional piece of hardware to seize the text and dialing in to a remote e-mail server. Our approach is to use regular phone handsets to dial to your home machine -supposedly up, and your voice modem monitoring incoming phone calls, will allow you to send an e-mail, and forward each other calls to an answering machine.
Introduction :
Phone based e-mail allows sending e-mail while you are away from your machine. Moreover, our way of approaching the problem bypasses the need of purchasing an additional piece of hardware to type in your messages. A regular phone
connected to the phone network will do. This convenience comes with the price of having your machine (at home) turned on. As PCs are becoming common everywhere for multipurpose use, this assumption does not seem too much demanding.
A typical scenario would be that you are away from your machine and want to send e-mail. Assuming that our Software is running on your machine, your voice modem will handle the incoming calls. Via a menu selection you will be prompted to either leave a message (usual phone calls), or to send an e-mail. The second option selected, you will be asked to enter a password, and than, voice directions will help you to know at each stage of the procedure of sending e-mail you are.
Program structure:
Our program acts as definite automata. Depending on triggering actions (phone ring) or phone keys entered, it changes its state (Fig -1). The initial state would be when your daemon is running on your machine, having set your modem to its voice mode. A phone ring will make the transition from the initial state to the "menu select" state. In other words, the modem picks up the call and a greeting message along with a voice menu is played. People calling you may leave a message when selecting to do so. The second option of sending email will be protected by a simple personnel password you have already set up.
The program goes on, with switching from state to state depending on the characters you enter. There are two distinctions to make here:
One is that while hearing a selection, a simple key strike-representing the number of your selection will change the state of your program. While the second is, inside the composing procedure, three key strikes, with the last as '*' means that you
Composed one character in your message/address; whereas striking '#', means finishing
Entering text and moving to the next step (either from composing message to composing email address or from composing email address to sending email).
Using the voice capabilities of the modem we are also able to play back each character
entered. Therefore the user can always go one step back, and correct the last character
entered. We used Voice Modem tools package [1] to explore the capability of playing back voice files through the modem.
Figure-1.
This figure represents the different states into which the program may be at a given time. State transition is operated when a phone's key is stroke, with the only exception at the initial state where transition is triggered by a phone ring.
The area in the dashed rectangle, represents the details of which could be just one
State, the message composition state.
Alphabet :
A major question would be, how can we code the Alphabet plus Numbers on such a key limited keyboard as the one of a phone's handset? Since each phone key has the label of three Letters, we alleviate the ambiguity of selecting one letter from the three by entering its sequential number in the key followed by the '*' key. For example on key #2, where the sequence 'ABC' appears, A has sequence number 1, B has 2 and C has 3. Therefore, to enter 'C', the user should strike the following keys: 2 , 3 , *. At a first glance such a schema appears to be very hard and long; however, the intuition underneath it is quite simple and obvious. Experience has shown that your hands will learn it quicker than you might think. (we experience it with students from different departments : humanities, management; and they seem to pick it up pretty quickly).
However, two optimizations has been included to reduce both the time of typing in Characters and reducing the risk of repetitive errors:
The first is a shortcut mechanism; some frequently used group of characters, will be referred to by combos similar to those coding a letter (Fig-2). For example "edu" will be entered by hitting 3 , 8 and "com" by hitting 2 , 6 (plus the "*" at the end).
|
Group of chars |
Phone keys |
Mnemonic |
|
edu |
3 ; 8 ; * |
E + U |
|
com |
2 ; 6 ; * |
C + O |
|
org |
6 ; 7 ;+ * |
O + R |
Figure -2
The mnemonic corresponds to the phone key where the letter occurs.
The second optimization is to have pre-written messages that we call "canned messages". In fact at each stage where you are susceptible of composing a text, you are first asked whether you would prefer to send out a canned message. Typical canned messages would be: < I am running late to the meeting>, <I am on my way home >,
<class is canceled today>, <call me up ASAP> . The same argument holds for an address
book. We also found it useful to have incomplete canned messages such as < I will arrive
at : >, and the user will be asked to enter numeric data for time.
After completing your message, hitting the "#" key after entering the e-mail address sends your message. A simple pipe is opened to the mail program in Linux, with the appropriate parameters: Address and subject on the line command plus message body.
Communication with the modem:
Through the serial port, we send AT commands [2]. That is the case for the initialization procedure as well as for setting the modem in the voice mode or enabling the modem to send or receive voice messages.
A Simultaneous Voice Data modem is characterized by its three modes:
This mode allows sending AT commands to the modem without aborting the telephone line connection.
This mode allows the recording of voice data, and is used when recording
a message for example.
Transmitting voice data is possible via this mode.
As far as our software is concerned, the communication with the modem was boiled down to two main problems:
The first one is, how to write data to the modem and how to read data from it. As stated before, we just write string of chars to the serial port where the modem is connected, and read from it string of chars. However, as obvious as it seems to be, this operation needed some tedious manipulation of Reverse Engineering techniques. The problem is mainly of synchronization, the modem sends sometimes some "bizarre" characters due to an non empty buffer, or due to noise in the line. Our approach is to keep reading from the buffer and throw away all the chars we judge as garbage. This is made possible by the state structure of the program: we know in advance what kind of information (alphanum, * or #) we are expecting to read from the buffer in each stage of the program.
The second one is the switch to Voice mode. The switch in itself is not the problem, this is done in a straight forward manner by sending the AT#CLS=8 command to the modem. Rather the problem was of figuring out the right sampling to use, along with the baud rate . We also explored the virtue of such mode that allows the sending of special chars to the modem in the stream and can be interpreted as a modem command. For example, the occurrence of the sequence <DLE><ETX> in a voice stream is interpreted as flushing the modem buffers. This property helped a lot for regularizing the behavior of the modem : Therefore, while playing a voice file through the modem, and appending to its end the sequence <DLE><ETX> we obtain a cleaning of its buffers at the same time.
Errors handling:
We suppose that the user has already prepared a message to send. A piece of paper containing the message to send should ease the confusion ones can have of which character to enter. In the other hand, as the sound of each character is played back, if by mistake the user enter a different character she should be able to go one step back and enter the correct character as explained in the previous section. However, if the source of error is just entering a code that do not match with any character in our alphabet, an error
message is played and the user can re-enter the code.
It might happen that the user would like to change a word he entered. A code number (as the ones entered for coding letters) will allow her to go back one word back (pointing the cursor in the buffer to the most recent space character entered). Moreover,
she might want to cancel out all the message and restart from the beginning. After the appropriate key combination is entered, the buffer is flushed and the cursor is repositioned at the beginning of the message.
Enhancements :
Including a reading e-mail component to the software seems to be a good idea, and will provide a full e-mail capability server. A synthesizer is needed or any other text-to-speech software. Such applications are already in use, and are mostly included in cellular phones, while the idea of sending e-mail via phone without the help of any other device seems to be new.
Conclusion :
SVD modems offer larger capabilities than regular modems. The additional,
Voice/Audio mode opens the door widely for new applications in Computer Telephony.
The idea of our software seems to be simple, however a lot of people are enthusiastic about it, and find it extremely useful. Many applications of our software in the industry come to mind: ISPs could offer Phone e-mail accounts for people who do not
have an e-mail account, or in the other hand it can be used by people who want to be able to send e-mail even if they are away from their machines.
References:
[1]. Voice Modem Tools downloaded from: ftp.vse.cz/pub/Linux/apps/serialcomm/!INDEX.html
[2]. Hayes modem AT commands:
http://www.computercraft.com/docs/hayescom.html