Queuing with distortion-control for multimedia content

There are numerous contexts in which systems need to buffer multimedia signal content. If such a buffer overflows, signal data are lost in an uncontrolled manner, which can lead to large end-to-end distortions. However, if the queued signals are distortion-tolerant, overflows can be avoided, and significant performance gains realized, by reducing the fidelity of the signals in a gradual, controlled, manner. Based on ideas of successive-approximation source coding, we design an adaptive-buffering algorithm to minimize end-to-end distortion. This algorithm performs nearly as well as a performance bound on all possible algorithms. Perhaps most importantly, the algorithm´s performance remains robust across a wide range of (unpredictable) utilization rates (input rate/output rate).