Terminal QoS of Adaptive Applications and its Analytical Computation
Pratyush Moghé, Asawaree Kalavade · 1997
Next-generation slim hosts or terminals (like PDA’s, settop boxes, and network computers) are expected to support sophisticated continuous media applications. Since these terminals have very limited processing power, the application processing delay can be a significant component of the end-to-end QoS. We call this delay a terminal QoS measure to distinguish it from end-to-end QoS measures like packet loss, delay, and delay jitter. Computing the processing delay can be tricky since many of the continuous media applications are smart and can adapt to network conditions. Such applications can use RTP/RTCP feedback of end-to-end QoS to continually change their output bitrate. Each adaptation is associated with a different amount of processing on the terminal. In this context, we pose the following problem. Consider a terminal that supports multiple concurrent adaptive applications. Each application uses end-to-end QoS feedback to adapt itself according to its adaptation algorithm. What effect do the adaptations have on the processing demand on the terminal? That is, what is the processing delay as a function of the adaptations? In this paper, we provide a theoretical framework to understand this interaction.