For determining the performance of ROHC, acticom offers a test and performance evaluation suite to analyze ROHC behavior for a wide range of link characteristics and application types. 

As depicted in the figure above, different quality levels must be distinguished: the network provider's view can be reflected by the network quality. The metric of the network quality is specified in RFC 3095 and describes an effective view of the achievable savings on transport capacity. The consumer's view is given by the objective/subjective video/voice quality and expressed by a PSNR/MOS value. The wireless link is modeled by traces obtained from GSM/DECT/WLAN measurements. The application generation utility provides easy creation of video/voice/data streams.

 

Our first measurements show the following savings of bandwidth used for acticom_ROHC using IPv4. The audio measurements (LPC, GSM, etc) were done with the robust audio tool (RAT) , while the video measurements are based on the H.26L codec (more information about H.26L).

 

For the investigation of voice and audio traffic we used the robust audio tool (RAT). RAT is a unicast and multicast audio conferencing tool and allows us to use different coding schemes (e.g. LPC, GSM, G.711) at different data rates. As an input for the RAT tool we used well-known sound files, converted into the WAV audio format. The WAV files are transmitted over the ROHC testbed using the RAT tool. At the receiver side the recorded soundfile was stored in WAV format. We wrote our own audiometer tool for quality comparison. The following table shows the result for the bandwidth savings using ROHC.

 

	upper bound	w/o UDP checksum	with UDP checksum
LPC	74%	71%	63%
GSM	55%	53%	47%

Using the videometer tool the impact of the video coding and the ROHC coding can be seen. The VideoMeter tool displays the original, the encoded, and the received YUV stream. Furthermore the VideoMeter tool also displays the differences between the YUV streams and plots the PSNR values over  time.