Marginal productivity index policies for scheduling multiclass delay-/loss-sensitive traffic

We address the problem of scheduling transmissions of heterogeneous packet traffic streams on a single link in the setting of a Markovian multiclass queueing model with losses, where every class has a dedicated finite buffer. Some classes correspond to real-time/delay-sensitive traffic (e.g. voice, video) whereas others correspond to nonreal-time/loss-sensitive traffic (e.g. data). Different levels of tolerance to delay and packet loss are modeled by suitable cost rates. The goal is to design well-grounded and tractable scheduling policies that nearly minimize the discounted or average expected cost objective. We develop new dynamic index policies, prescribing to give higher service priority to classes with larger index values, where the priority index of a class measures the marginal productivity of work at its current state. Such index policies are shown to furnish new, insightful structural results.