VoIPTool 2.0
============

This is the OMNeT++ 4.x port and complete rewrite (by Zoltan Bojthe)
of the TKN VoIPTool, http://www.tkn.tu-berlin.de/research/omnetVoipTool/

The fraction of voice over Internet Protocol (VoIP) based telephone calls
among the totality of voice based communication acts has been significantly
growing during the last years. In wired as well as wireless communication
applications, VoIP is expected to completely replace former circuit
switched telephony. This project provides an OMNeT++-based VoIP traffic
generator that creates realistic VoIP packet streams, due to the
utilization of real audio data and an existing VoIP standard codec.
Moreover, by applying ITU-T's perceptual evaluation of speech quality
(PESQ) approach at the sink, the perceived quality of a transmitted VoIP
stream can be determined.

This library is an add-on to the INET Framework, http://inet.omnetpp.org.
It provides two modules: VoIPSourceApp and VoIPSinkApp. Both are
application- layer modules that operate over UDP, and can be used in INET's
StandardHost very much like other UDP traffic generators and sinks (see the
udpApp[] submodule vector, numUdpApps parameter, etc. in StandardHost.)

VoIPSourceApp accepts an audio file and a destination IP address/port as
input, and will transmit the file's contents as voice traffic over UDP n
times (by default once). For transmission, the audio is resampled at the
given frequency (by default 8KHz) and depth (by default 16 bits), and
encoded with the given codec (by default G.726) at the given bit rate (by
default 40Kbps), and chopped into packets that each carry dt milliseconds
of voice (by default 20ms). (The above default values may be out of date;
actual values come from the NED file).  Packets that are all silence (all
samples are below a given threshold in absolute value) are transmitted as
special "silence" packets. The module does not simulate any particular VoIP
protocol (e.g. RTP), but instead accepts a "header size" parameter that
can be set accordingly.

VoIPSinkApp listens on an UDP port, and expects to receive VoIP packets on
it. The received voice is then saved into a result audio file that can be
compared with the original for further evaluation. VoIP packets are
numbered, and out-of-order packets are discarded (the corresponding voice
interval will be recorded as silence into the file). VoIP packets that miss
their deadlines will similarly be discarded. It is assumed that the audio is
played back with delay (by default 20ms), which allows some jitter for the
incoming packets. The resulting audio file is closed when the simulation
completes (i.e. in the OMNeT++ finish() function).

Related publication:

M. Bohge and M. Renwanz, "A realistic VoIP traffic generation and evaluation
tool for OMNeT++", 1st International OMNeT++ Workshop, March 2008, Marseille,
France.


---
Andras