CALL ADMISSION FOR PRERECORDED SOURCES WITH PACKET LOSS

We develop call admission policies for statistically multiplexing prer ecorded sources over a bufferless transmission link. Our model is appr opriate for video on demand, as well as other on-demand multimedia app lications. In particular, we allow users to specify when the sources b egin transmission; we also allow the user to invoke VCR actions such a s pause and temporal jumps. We suppose that the quality of service (Qo S) requirement allows for a small amount of packet loss. We develop a stochastic model which captures the random phases of the sources. We t hen apply large deviation theory to our model to develop global admiss ion rules. The accuracy of the large deviation approximation is verifi ed with simulation experiments employing importance sampling technique s. We also propose a refined admission rule which combines the global test and a myopic test. Numerical results are presented for the Star W ars trace; we find that the statistical multiplexing gain is potential ly high and often insensitive to the QoS parameter. Finally, we develo p efficient schemes for the real-time implementation of our global tes t. In particular, we demonstrate that the Taylor series expansion of t he logarithmic moment generating function of the frame size distributi on allows for fast and accurate admission decisions.