THE PROPOSED LANGUAGE EXTENSIONS

In this section a set of language extensions is described that allows the multimedia author to add QoS and real-time requirements on the media that participate in multimedia applications. The language that will be extended is SMIL [3], a language for Web-based multimedia presentations which has been developed by the W3C working group on synchronized multimedia. In this section we introduce and define a new aspect of a multimedia presentation, called quality specifications. This set of primitives that extends the basic SMIL language is used to express the required QoS for all the continuous media objects that participate in any presentation. We assume that the QoS for continuous media objects is expressed with temporal and spatial resolutions. The temporal resolution can be expressed by the number of frames per second or sample rate and the spatial resolution can by expressed by data size or number of bits per pixel.

SMIL supports two types of continuous media objects, video and audio. In our extended SMIL language, these two objects can be described together with their quality requirements within a document via the following syntax:

• $\<$video cmo-attributes v-qos-attributes$\>$, and
• $\<$audio cmo-attributes a-qos-attributes$\>$.

The proposed extensions are defined by the two new sets of attributes v-qos-attributes and a-qos-attributes for video and audio respectively. The set cmo-attributes (cmo stands for continuous media object) is curenlty supported by SMIL and it is used to define the location of the media object, its duration or the number of times this object should be repeated. The new v-qos-attributes list describes video quality requirements using the attributes:

fps

: The value of fps defines the temporal resolution of a video presentation by giving the number of frames per second. For example giving fps=14-18 as attribute to a video object, it means that the accepted values for this video presentation can be any rate between 14 and 18 frames per second (Figure 2 lines: L-11,L-12,R-10).

spatial-res

: The spatial-res definition of a video presentation specifies the spatial resolution in pixels required for displaying the video object. (Figure 2 lines: L-11,L-12,R-10).

color-res

: This attribute specifies the color resolution in bits required for displaying the video object. During object presentation the highest color resolution is tried to be used first and it is decided at run time according to the availability of system resources (lines: L-11,L-12,R-10).

The list a-qos-attributes is the last set of SMIL extensions and describes audio quality requirements (lines: L-13,R-11), using the attributes:

sample-rate

: The value of sample-rate defines in KHz the rate that the analog signal is sampled.

sample-size

: This attributes specifies the sample size in bits of each sample. The highest value that can be used for every sample it is decided at run time according to the availability of the resources.

The above language primitives form a complete set for QoS definition of every distinct continuous media that participate in a multimedia presentation. If several media streams have to be combined then inter-media synchronization is another important factor of quality specification but this subject has been extensively studied and completely supported by the standard SMIL language.