How to try out SEMS without setting up a proxy

Introduction

This text describes how one can try out services in SEMS without setting up a proxy. This is the simplest way to try services in SEMS, or start with developing a service.

The way this works is that SEMS registers to a SIP server (registrar) with one account (bob), just like any other SIP phone, and we call SEMS from another account (alice). If the SIP server provides DID calling from the PSTN, we can use any landline or mobile phone for testing or using the service.

Note:

What is not possible with this method is to use applications which need additional information for a call from the subscriber data. For example, in order to send a voicemail as email the SEMS server needs the email address to send the mail to.

Requirements

For compiling SEMS, as a minimum a C++ compiler and make is needed. In debian, do

 apt-get install g++ make 

With a public SIP server

Two accounts at a public SIP server are needed. We recommend to use iptel.org's SIP service for testing, an account can be registered for free at http://iptel.org/service/.

Any SIP phone, hardphone or softphone, can be used for testing. Cross-platform, sip-communicator is recommended, for Linux twinkle, for Windows sip-communicator, xten eyebeam or NCH express.

With a PSTN DID provider

Alternatively a PSTN DID provider can be used. In that case, we can test and use the service with any phone. A list of DID providers is for example available at voip-info . Sipgate for example, provides free DID numbers in Germany.

Installing SEMS from source

First, the SEMS source is downloaded from iptel.org and extracted:

     $ wget ftp.iptel.org/pub/sems/sems-latest.tar.gz
     $ tar xzvf sems-latest.tar.gz

SEMS is compiled:

    $ cd sems-x.y.z/
    $ make

Note:

Compilation may fail for some modules due to missing dependencies. For most modules, that can be ignored for the moment.

Then SEMS is installed:

    $ make install

This will install

• configuration in /usr/local/etc/sems/
• the sems binary in /usr/local/sbin/sems
• modules in /usr/local/lib/sems/plug-in/
• audio files in /usr/local/lib/sems/audio/

Configuring the application for SEMS

There are many many modules shipped with SEMS, applications like announcement, voicemail, conference, etc, codec modules, and some things like SIP registrar client.

Now we configure SEMS to load the conference application and execute the conference application for incoming calls. We also set it to have itself register to our SIP server.

In /usr/local/etc/sems/sems.conf, we set

   load_plugins=wav;uac_auth;registrar_client;reg_agent;conference   

to load the modules we need; wav is for reading WAV files and for the G711 codec, uac_auth is the module which implements authentication, registrar_client facilitates registration at a SIP server, and reg_agent is the application that uses registrar_client to have SEMS register at a SIP server.

We also set

   application=conference

so that SEMS executes the conference application for an incoming call.

We want SEMS to register at a SIP server, so we need to tell it about the user name and the password, this is set in /usr/local/etc/sems/reg_agent.conf (of course this user name bob and the password need to be set to the ones used for testing):

     domain=iptel.org
     user=bob
     display_name=bob
     auth_user=bob
     pwd=verysecret

Testing the setup

Now we can test the configuration by running SEMS from the command line like this:

    /usr/local/sbin/sems -f /usr/local/etc/sems/sems.conf -D 3 -E

-D 3 sets the debug level higher so that we see what is going on, and -E makes SEMS start in the foreground and go to daemon mode. It also makes the log appear on the terminal and not in the system log file.

If everything is alright, SEMS starts up with a lot of messages, and hopefully no ERROR. There should also be some messages appearing which show that SEMS registered successfully to the SIP server.

Now we can call bob from the other phone or our PSTN telephone. In the SEMS log, we see the call appearing, and on the phone we hear a message saying that we are the first participant in the conference.

If it doesnt work.... we examine the log for the ERROR that occured. Possibly, depending on the network setup, we need to change the interface that SEMS is running on; this can be changed by setting the media_ip and sip_ip options in sems.conf. Also, it might be that there is already someone using that port (default config: 5070), in that case sip_port needs to be set.

Running as daemon

If SEMS is started without the -E option, it will continue running as daemon in the background. The log can be seen in syslog (e.g. with tail -f /var/log/daemon.log).

Running other applications

If we want to run other applications, the load_plugins= and application= parameters need to be adapted. See Application Modules for a description of the shipped applications.

Creating and running a simple DSM applications

The DSM is a service development platform, that makes it simple to create powerful services. The service logis is defined as a state machine, and the DSM application interprets this state machine for the calls, evaluating when to change state, and which actions to execute.

To use a DSM application, we set in /usr/local/etc/sems/sems.conf

   load_plugins=wav;uac_auth;registrar_client;reg_agent;session_timer;dsm
   application=mydsmapp

and in /usr/local/etc/sems/dsm.conf :

    diag_path=/usr/local/lib/sems/dsm/
    load_diags=mydsmapp
    register_apps=mydsmapp

Then we paste this little script in /usr/local/lib/sems/dsm/mydsmapp.dsm :

  initial state BEGIN
   enter {
     playFile(/usr/local/lib/sems/audio/webconference/first_participant.wav);
   };
  transition "file ends" BEGIN - noAudioTest -> TYPING;
  
  state TYPING;
  transition "typed a key" TYPING - keyTest(#key < 10) / {
    set($myfile=/usr/local/lib/sems/audio/webconference/);
    append($myfile, #key);
    append($myfile, .wav);
    playFile($myfile);
  } -> TYPING;

  transition "BYE received" (BEGIN, TYPING) - hangup / stop(false) -> END;
  state END; 

This little script welcomes the caller, and then plays the key that the caller entered. More documentation about DSM and examples are in apps/dsm/doc, and also DSM: State machine notation for VoIP applications .