Conclusions

This Chapter presented a detailed description of our new method which aims to assess real-time multimedia streams transmitted over packet networks in real time. We presented the methodology in general regardless of the media type. Our method can be used for speech, audio, video or multimedia quality assessment. Session types also can be taken into account (one-way, interactive or multiparty conferences). The development of the method consists of the following steps. First, the quality-affecting parameters should be identified. Then a set of sequences distorted by wide range variation of these parameters should be constructed. After that, a suitable subjective quality test is performed on these sequences. The result is a database consisting of a set of parameters' values and the corresponding quality scores. A suitable neural network architecture should be identified, trained and tested by the resulting database. The result is a neural network tool that can be used to assess in real time the quality of the chosen media when transmitted in real time over packet networks. One of the advantages of our method is that it takes into account the direct effect of network parameters (e.g. loss rate, jitter, etc.) and the codec parameters (e.g. bit rate, frame rate, etc.), as well as the other hybrid parameters (e.g. echo, crosstalk, etc.). In addition, it is a hybrid method between subjective quality measures (human perception of the quality) and objective measures (it can work in real time). This Chapter presented some of the advantages of using our method. We also provided a guideline about how to carry out the suitable subjective quality tests, which constitutes the most difficult part in implementing that method. In addition, the interaction between our method and the existing real time network components in operation mode is provided. Validation of the methodology is postponed to the next two Chapters in which, we verify the applicability of it to correctly evaluate in real time the quality of speech and video transmitted in real time over IP networks. Based on these results, we provide a study of the impact of the quality-affecting parameters on the quality in Chapter 7. Based on that study and that method, we give in Chapter 8 a new rate control protocol that takes into consideration not only the traditional network parameters measures for the rate control, but also the user's perception of the quality to dynamically change the sending parameters in order to guarantee the best possible quality.