Demo for the paper: Kwasinski and N. Farvardin, ``Extending the Operation
of a CDMA Network Beyond Congestion by Real-time Source Adaptation: Analysis and
Design'', submitted to IEEE Transactions on Communications.
This demo shows different speech samples to compare the difference in performance between the rate-adapted systems and the systems with no adaptation as described in the paper.
The best way to go through this demo is to play the different samples (original and received) following the curves for the adapted system as the normalized number of users (and, consequently the normalized distortion) increases. Here "Original" stands for the original sample and "Received" is the sample after going through all the communication system. For each pair of samples, original and received, there are one or two samples that have approximately the same distortion but that corresponds to the behavior of the non-adapted systems (this is shown by the dashed arrow for the case of distortion about 1.4). In this way it is possible also to compare the behavior of both systems (adapted and non-adapted) and see how the difference in operation, the adapted increasing distortion by source rate reduction and the non-adapted by increasing channel-induced distortion, affects differently the performance. In many cases there are two samples for each distortion value of the non-adapted system. This is to show how the growth in distortion due to the increase in channel-induced errors may generate results that have the same objective distortion measure but are perceived differently.
In the paper, we set the limit on acceptable distortion at about 1.4. Some of the samples here that exceeds that distortion value are still intelligible, yet we consider that in most of the cases they are not acceptable in terms of perceived quality for a realistic commercial communication system.
To play the samples, simply click on the corresponding speaker icon on the figure.