Optimal multiplexing of packetized traffic with different requirements

The authors consider the problem of finding a transmission channel management policy, such that the average delay of some special packetized traffic is bounded while the weighted sum of average delays of the remaining traffic is minimized. An efficient algorithm is given to find an optimal policy based on the randomization of a subset of all fixed priority order taken with the right probability. Easy-to-check conditions are also given for feasibility and global optimality in the family of all work conserving policies. The results can be of interest in many dimensioning and management problems that can be fund in multiservice packetized networks.<
>